• Get 30% Off with code "NOV"
  • 40% OFF GROWTH ACCOUNTS
  • ENDS 13 NOVEMBER, 11:59PM EST

How to Build a Trading Strategy That Survives a Test

An overhead desk with charts on two screens as a trader works, under the headline How to Build a Trading Strategy.

TL;DR: How to build a trading strategy comes down to writing rules a stranger could follow, then testing them on data you have not seen. In a 5-minute ES pullback test, the raw idea won 31.51% of 146 trades against a 36.13% break-even win rate. Judge win rate together with payoff, change one rule at a time, and keep size small until the trade count can carry a verdict.

Most traders who want a strategy already have an idea. They like buying pullbacks, or fading the open, or trading breakouts after the first hour. The hard part is not the idea. It is turning it into something you can measure, so that a bad week tells you something instead of just hurting.

The usual reaction to a rough week is to add a filter or nudge the stop. Sometimes that helps. More often it just changes which trades you took, and you never find out whether the original idea had anything in it.

This guide builds one plain idea into a full rule set, then tests it on real E-mini S&P 500 and E-mini Nasdaq-100 bars with a sealed out-of-sample window. The result is not flattering, and that is the point. A strategy process is worth having because it tells you when an idea does not work before your account does.

What a trading strategy is and what it is not

A trading strategy is a written rule set that answers six questions: what you trade, when you may act, what triggers the entry, where you admit you are wrong, what ends the trade, and how much you risk. The test is simple. If a competent stranger read your rules, would they take the same trades you take? If not, you have a preference, and a preference cannot be measured or improved on purpose.

Two things often get called a strategy that are not one.

  • A setup is a pattern you recognize on a chart. It has no rule for size, no rule for the stop, and no rule for what happens when it shows up while you are already in a position.
  • A trading plan sits one level above the strategy. It covers your capital, your daily loss limit, your schedule and your review routine. The plan says how much you may lose. The strategy says which trades you take to lose it on.

A trading strategy also needs a reason to exist. Trend continuation, mean reversion, breakouts and range fades are different answers to the same six questions. None of them is better in the abstract. Each one becomes testable only when it is written down that way.

Related lesson: How to create a trading plan

Setting goals and choosing a trading style for the strategy

Strategy design starts with constraints, not with charts. How many hours can you actually watch a screen? What would you consider a good outcome after three months? How much can you afford to lose while you find out? Your answers narrow the trading style before you write a single entry rule.

Goals should describe process, not money. "Follow the written rules on every signal for 60 sessions and log every trade" is a goal you control. "Make $500 a day" is an outcome that depends on edge, size and luck, and it quietly pushes you toward bigger size whenever you are behind.

The main trading styles trade off frequency against holding risk:

Style Typical chart Holding time What it costs you
Scalping 1-minute or tick charts Seconds to minutes Costs are large relative to each target, and execution quality decides the result
Day trading 5-minute to 15-minute charts Minutes to hours, flat by the close Needs screen time during the session you trade
Swing trading 1-hour to daily charts Days to weeks Carries overnight and weekend gap risk, and produces few trades to measure
Position trading Daily to weekly charts Weeks to months Very few trades per year, so validation takes years

The timeframe choice also decides how fast you can learn. A strategy on 5-minute bars of the E-mini S&P 500 during regular hours (9:30 AM to 4:00 PM ET) can produce more than a hundred signals in a couple of months. A daily-bar strategy on the same market might produce fewer than twenty a year. If you can only test what you have time to trade, choose the timeframe from your hours first, then keep every test on that timeframe.

Related lesson: Strategy timeframes

Choosing the market to trade

The market you pick is part of the strategy, because the same rule behaves differently on a different instrument. Four questions decide whether a market fits:

  • Liquidity. Can you get filled near the price you expect at the times you trade?
  • Hours. Is the market open and active when you are free? CME Globex equity index futures trade from Sunday 6:00 PM to Friday 5:00 PM ET with a daily break from 5:00 PM to 6:00 PM ET, but most volume clusters around the 9:30 AM ET cash open.
  • Contract size. Can you express your stop at a dollar risk you can afford?
  • Cost relative to the stop. How much of each trade's risk goes to commission and slippage?

Contract size is where most new strategies quietly break. The E-mini S&P 500 (ES) is worth $50 per index point, and its minimum tick of 0.25 points is worth $12.50. The Micro E-mini S&P 500 (MES) is worth $5 per point, one tenth of ES. The E-mini Nasdaq-100 (NQ) is worth $20 per point, with a 0.25-point tick worth $5.00.

Here is the math with round, illustrative numbers. Say your rule wants a 7-point stop on the S&P 500. On ES that is 7 x $50 = $350 of risk per contract. On MES it is 7 x $5 = $35. If your risk budget per trade is $100, ES cannot express the trade at all, while MES lets you take two contracts for $70 of risk. The strategy did not change. The market choice decided whether it was tradable at your size.

Stocks, spot forex and crypto each answer these questions differently. Stocks offer thousands of symbols but trade mostly during exchange hours. Spot forex trades around the clock on weekdays. Crypto trades every day. Whichever you pick, start with one instrument, learn its rhythm, and only add a second one once the first has a measured record.

Related lesson: How to choose a market to trade

Technical and fundamental analysis as rule inputs

Every strategy acts on evidence. The question is which kind, and whether you can turn it into a condition a computer or a stranger could check.

Technical analysis uses price, volume and indicators derived from them. "Price closed above the 20-period exponential moving average" is a technical condition. So is "the 14-period ATR is above its 50-bar median." Technical rules are the easiest to test because the inputs sit in your chart history.

Fundamental analysis uses information about the thing being traded: earnings, economic data, interest rates, supply and demand. For futures day traders, the most practical fundamental input is the economic calendar. The Bureau of Labor Statistics releases the Consumer Price Index at 8:30 AM ET, and index futures often move sharply in the minutes after it. A fundamental rule might be "no new entries in the 15 minutes before or after a scheduled CPI release." That is still a testable condition.

Quantitative rules sit between the two. They use statistics, such as volatility percentiles or return distributions, to decide when to act. Most testable strategies end up blending all three: a technical trigger, a volatility filter, and a calendar rule that keeps you out of known event risk.

The trap is mixing them without writing it down. "I take the pullback unless the macro backdrop looks bad" is a discretionary override. That can be legitimate, but it has to be recorded as its own rule, or your results will mix two different strategies and you will never know which one worked.

Defining the trend in a testable strategy

Many strategy ideas start with "in an uptrend." That phrase is not a rule until you define it. Three common ways to make it mechanical:

  • Stacked moving averages. The 20-period average is above the 50, and the 50 is above the 200. Simple to test and impossible to argue with after the fact.
  • Market structure. Price has made a higher high and a higher low on your chosen timeframe. This needs a rule for what counts as a swing point, or two traders will mark different swings.
  • Trend strength indicators. ADX above a threshold says the market is trending without saying which way, so it is usually paired with a direction rule.

Each definition changes which trades you take. A stacked-average filter reacts slowly and keeps you out of early reversals. A structure rule reacts faster but needs a precise swing definition. Neither is correct. What matters is that you choose one, write it down, and test it as a separate decision from the entry trigger.

For the test used throughout this guide, "uptrend" means the 20, 50 and 200-period exponential moving averages on 5-minute bars are stacked in that order. An exponential moving average weights recent prices more heavily than older ones, so it turns faster than a simple average of the same length.

Related lesson: Market structure

Entry rules a stranger could follow

The three parts of an entry rule a stranger could follow: the trigger, the location where it counts and the fill assumption.

An entry rule needs three parts: the trigger, the location, and the fill assumption.

  • Trigger. The exact event that creates a signal, such as "a 5-minute bar closes above the 20-period EMA after the previous bar closed at or below it."
  • Location. Where that trigger counts. "Only when the 20, 50 and 200 EMAs are stacked upward" is a location rule.
  • Fill assumption. When and at what price you assume you got in. "At the open of the next bar" is honest. "At the close of the signal bar" assumes you can act before the bar finishes, which you cannot.

Strike words that need judgement. "Strong close," "clean pullback," "obvious level" and "valid setup" are opinions. If a word in your entry rule would make two traders disagree, replace it with a number or a structural definition.

The entry rule for the ES and NQ pullback strategy tested in this guide is: go long at the open of the next 5-minute bar after a bar closes back above the 20-period EMA, during regular trading hours, with one position at a time. No discretion, no second timeframe.

Also decide what happens when a signal fires while you are already in a trade. Ignore it, add to the position, or reverse? Most traders never write this down, and it changes results more than they expect.

Related lesson: Entry and exit strategies

Stop and exit rules for the strategy

The stop is where the strategy admits it was wrong. The exit is where it takes what it came for. Both need rules as specific as the entry.

Common stop methods:

  • Structural stop. Beyond the swing low or high that the trade depends on. Logical, but the distance varies a lot from trade to trade.
  • Volatility stop. A multiple of the Average True Range (ATR), which measures the typical bar range including gaps. The stop widens on busy days and tightens on quiet ones.
  • Fixed stop. A set number of points or ticks. Easy to size, but blind to conditions.

Common exits:

  • Fixed target. A set multiple of the risk, such as 2R (twice the stop distance).
  • Trailing exit. The stop follows price after a gain, so winners can run.
  • Time exit. Close the trade after a set number of bars or at a set time, such as the session close.

The ES and NQ pullback strategy uses a volatility stop and a fixed target: the stop sits 1 x the 14-period ATR below entry, the target sits 2 x ATR above, and any trade still open at the 4:00 PM ET close exits at the last bar's close. ATR here is Wilder's 14-period version, calculated on regular-hours 5-minute bars.

The exit rule matters as much as the entry. The same trigger with a different stop and target is a different strategy with a different win rate, a different payoff and a different break-even point. Change it on purpose and test it as its own version.

Related lesson: Risk-reward ratio

Risk management and position sizing in a strategy

Position sizing turns a stop distance into a dollar risk you chose in advance. The formula for futures is:

Contracts = risk budget per trade ÷ (stop distance in points x dollars per point), rounded down.

A worked example with round, illustrative numbers. Your risk budget is $200 per trade and the rule's ATR stop is 6 points on the S&P 500.

  • On ES: 6 x $50 = $300 per contract. $200 ÷ $300 = 0.67, which rounds down to zero. The trade is too big for your budget.
  • On MES: 6 x $5 = $30 per contract. $200 ÷ $30 = 6.67, which rounds down to 6 contracts and $180 of actual risk.

Round down, never up. Rounding up turns a $200 budget into a $210 or $240 risk, and those extra dollars compound across a losing streak.

Losing streaks are not rare. A strategy that wins 35% of the time loses about 65% of the time, and the chance that any given run of five trades is all losers is 0.65 to the fifth power, about 11.6%, using simple probability and assuming each trade is independent. Over a few hundred trades you should expect several five-trade losing runs and at least one much longer. Size so that the worst streak you can reasonably expect leaves you able to keep trading the rules.

This is where many strategies actually fail. Not in the math, but on loss number six, when the trader starts skipping signals or doubling size to get it back. A size that feels slightly too small on a good day is usually the right size on a bad one.

Two more risk rules belong in the strategy itself:

  • Aggregate exposure. If you trade ES and NQ together, they usually move in the same direction. Two long positions are closer to one double-sized bet than two separate ones.
  • A daily stop. A maximum daily loss, after which you stop for the session, keeps one bad day from turning into a bad month.

Related lesson: Position sizing

Writing the strategy down as a one-page spec

A one-page strategy spec writes down every decision, market, session, entry, stop, exit and size, with a version number and no discretion.

Writing a strategy down does two jobs. It forces every decision into the open, and it gives you a fixed version to test, so you can tell later whether a change helped or whether you just remember it differently.

Here is the spec for the pullback strategy tested in this guide, filled in:

Decision Rule for version 2
Market E-mini S&P 500 (ES), front-month continuous series; E-mini Nasdaq-100 (NQ) as a second check
Session Regular trading hours, 9:30 AM to 4:00 PM ET, 5-minute bars
Trend definition 20, 50 and 200-period EMAs stacked in that order
Entry trigger Bar closes above the 20 EMA after the prior bar closed at or below it; enter at the next bar's open
Stop 1 x 14-period ATR below entry, rounded to the nearest tick
Exit Target 2 x ATR above entry, otherwise exit at the 4:00 PM ET close
Position limit One position at a time, long only
Cost assumption $5 round turn plus one tick of slippage per trade
Discretion None

Keep a version number at the top and a change log underneath. Version 1 is the raw idea. Version 2 adds the trend filter. Version 3 adds a volatility filter and a time cutoff. When you look back after 100 trades, you need to know exactly which rules produced which results.

The discretion field is not a weakness. If you allow judgement, say where it lives ("may skip signals in the 15 minutes before a scheduled CPI release") so your log can separate the trades you took by rule from the ones you took by feel.

How to test a trading strategy you have built

Testing a trading strategy means running the frozen rules over historical data and reading the result honestly. The most important step happens before you look at any results: split the data by date and seal the later part.

The ES and NQ pullback test design

The test used the three versions of the pullback strategy on 5-minute bars of the continuous front-month E-mini S&P 500 and E-mini Nasdaq-100 futures, regular trading hours only, from August 28 to November 4, 2026. The data came from Yahoo Finance's chart service.

  • Sessions: 46 per market. September 14 and 15 were removed because the continuous series switched from the September to the December contract mid-session. One corrupted 3:00 PM bar on October 30 was removed.
  • Warm-up: the first 200 bars were used only to start the moving averages, so trades begin on September 1.
  • In-sample window: the first 28 sessions, August 28 to October 9, where rules would be written.
  • Out-of-sample window: the last 18 sessions, October 12 to November 4, which were only read once.
  • Cost: an assumed $5 round turn for commission and fees plus one tick of slippage on every trade. That is 0.35 points ($17.50) per trade on ES and 0.50 points ($10.00) on NQ.

Version 1 is the raw idea: buy the close back above the 20 EMA, no filter. Version 2 adds the stacked-average trend filter. Version 3 adds two more conditions: the 14-period ATR must be above its own 50-bar median, and no new trade may open after 2:00 PM ET.

What the three strategy versions returned

Results across the whole window, per contract, net of the cost assumption:

Version ES trades ES win rate ES payoff ES break-even win rate ES average net points NQ trades NQ average net points
1, raw idea 146 31.51% 1.77 36.13% -0.85 132 -3.16
2, trend filter 60 28.33% 1.73 36.63% -1.37 56 -6.84
3, trend plus volatility and time filters 17 41.18% 2.20 31.22% +2.11 18 -0.82

Payoff is the average net winner divided by the average net loser. The break-even win rate is 1 ÷ (1 + payoff), the win rate the strategy needs just to break even at that payoff.

On ES, version 1 lost $6,192 per contract and version 2 lost $4,125. Version 3 made $1,790 on 17 trades. On NQ, all three versions lost: $8,345, $7,660 and $295 per contract.

Reading the in-sample and out-of-sample split

Version 3 looks like the answer until you split it by window:

ES version In-sample trades In-sample win rate In-sample average net points Out-of-sample trades Out-of-sample win rate Out-of-sample average net points
1 79 27.85% -1.45 67 35.82% -0.14
2 33 24.24% -2.27 27 33.33% -0.28
3 7 28.57% -1.03 10 50.00% +4.30

In the window where the rules would have been written, version 3 lost money on 7 trades. Its whole positive result comes from 10 trades in the sealed window. The 95% confidence interval on its out-of-sample average runs from -3.69 to +12.29 points, which comfortably includes zero, and its t-statistic is 1.05.

Version 3 across the full window does no better on that test. Its average of +2.11 points has a standard deviation of 11.53 points across 17 trades, giving a 95% interval from -3.38 to +7.59 points. On NQ the same rules averaged -0.82 points.

Here is the important part. The honest verdict is that none of the three versions separates from zero in this sample. That is a useful result. It tells you this idea, as written, has no measured edge, and that the next step is a different idea or a much bigger sample, not more filters.

Related lesson: Backtesting a trading strategy

One trade resolved bar by bar

Aggregate tables hide what a pullback strategy feels like to trade. Two trades from the out-of-sample window of the ES pullback test (version 2 rules, 5-minute bars) show it.

On October 28, 2026, the bar stamped 9:50 AM closed back above the 20 EMA with the averages stacked. ATR was 6.08 points. Entry was the next bar's open at 7,763.25, the stop sat at 7,757.25 and the target at 7,775.50.

Bar (ET) Open High Low Close Low minus stop
9:55 AM 7,763.25 7,766.50 7,761.25 7,765.75 4.00
10:15 AM 7,769.25 7,770.75 7,765.75 7,770.75 8.50
10:35 AM 7,768.50 7,769.00 7,762.50 7,763.25 5.25
10:55 AM 7,761.75 7,763.75 7,759.25 7,762.25 2.00
11:20 AM 7,771.00 7,775.25 7,771.00 7,773.25 13.75
11:35 AM 7,772.25 7,775.50 7,770.50 7,775.00 13.25

The trade hit its target on the 11:35 AM bar, 21 bars after the signal, for 12.25 points gross and 11.90 net, or $595 per ES contract. Look at the 10:55 AM row. The eventual winner traded 2.00 points, eight ticks, above its stop after more than an hour. That is the moment a trader moves the stop, cuts the trade early, or holds to the rule. The outcome could not tell you in advance which choice would be rewarded.

The opposite case came on October 20, 2026. The bar stamped 1:20 PM gave a signal with ATR at 7.45 points. Entry was 7,761.75, stop 7,754.25 and target 7,776.75. Price rose to 7,768.50 on the 1:50 PM bar, 6.75 points in profit and 8.25 points short of the target, then drifted lower for 45 minutes and hit the stop on the 2:35 PM bar. The result was -7.85 points net, or -$392.50 per contract.

Same rules, same market, same month, opposite outcomes, and neither trade says anything about the other. That is why a handful of trades cannot carry a decision about a strategy.

Tracking strategy performance with the right numbers

Strategy performance tracking starts with one habit: record every trade the rules produce, not only the ones you remember. Then read the numbers together, because each one alone misleads.

Strategy win rate only means something next to payoff

A win rate means nothing until you pair it with the payoff it was earned at. The break-even win rate is:

Break-even win rate = 1 ÷ (1 + payoff ratio)

In the ES pullback test, version 1 won 31.51% of its 146 trades with a net payoff of 1.77. Its break-even win rate was 1 ÷ 2.77, or 36.13%. It fell 4.62 percentage points short, so it lost money even though nearly a third of trades won.

Cost moves that line. The 0.35 points of cost per ES trade looks tiny, but against an average risk of 6.70 points it is about 5.2% of every trade's risk. Before cost, version 1 averaged -0.50 points per trade, so cost made a losing idea worse rather than turning a winner into a loser. On a strategy with a thin edge, that 5% is often the whole edge.

Expectancy ties the pieces into one number: average net result per trade, in points or dollars. It already includes win rate, payoff and cost, which is why it is the number to track once the sample is large enough to trust.

The trading journal fields that make review possible

Track these fields for every trade:

  • Strategy version and the exact signal time
  • Entry, stop and target prices, and the actual fill
  • Exit price, exit reason (stop, target, time) and net result after cost
  • Maximum adverse excursion, the worst point the trade reached against you
  • Maximum favorable excursion, the best point it reached in your favor
  • Whether you followed the rule, and if not, what you did instead

Maximum adverse excursion is the column traders feel and rarely record. The October 28, 2026 ES winner came within 2.00 points of its stop. If your log shows many winners that come that close, your stop is tight for this market, and that is a measured reason to test a wider one as a new version.

Related lesson: How to keep a trading journal

Optimizing strategy settings without curve fitting

Optimization means changing the strategy's settings to improve its results. Done badly, it finds the noise in one sample and calls it an edge. The ES pullback test shows how quickly that happens.

The test ran the version 2 rules of the ES pullback strategy (buy the next 5-minute bar after a close back above the 20 EMA, with the 20, 50 and 200 EMAs stacked upward) with 25 combinations of stop and target: stops of 0.5, 1.0, 1.5, 2.0 and 2.5 x ATR against targets of 1, 2, 3, 4 and 5 x ATR. Each cell was measured in the in-sample window (August 28 to October 9, 2026) and then read once in the out-of-sample window (October 12 to November 4, 2026). Values are average net points per trade, in-sample first.

Stop (x ATR) Target 1 x ATR Target 2 x ATR Target 3 x ATR Target 4 x ATR Target 5 x ATR
0.5 -1.47 then +0.37 -1.15 then -0.05 -0.99 then +0.02 -1.18 then -0.25 -2.30 then +0.14
1.0 -2.13 then +0.23 -2.27 then -0.28 -2.33 then -0.18 -2.07 then +1.11 -3.92 then +0.93
1.5 -1.54 then +0.23 -1.69 then -1.20 -2.33 then -1.44 -2.67 then +0.01 -2.96 then +1.75
2.0 -1.38 then +1.31 -2.31 then +0.94 -2.71 then +1.33 -3.06 then +1.10 -3.15 then +3.52
2.5 -1.75 then +3.54 -1.48 then +3.21 -1.42 then +4.38 -1.19 then +4.77 -0.26 then +4.95

What the strategy grid shows in and out of sample

In the ES pullback grid, all 25 cells lost money in-sample. The median cell lost 2.07 points per trade. A trader following the process would have stopped here and moved on. That is harder than it sounds, because 25 losing cells make the next tweak feel overdue.

Out-of-sample, 19 of the 25 cells made money, with a median of +0.93 points. The same rule set flipped sign between two windows a few weeks apart. The best in-sample cell (2.5 x ATR stop, 5 x ATR target) lost 0.26 points in-sample and made 4.95 out-of-sample, on only 14 trades. The rank correlation between in-sample and out-of-sample results across the 25 cells was 0.07, close to no relationship at all.

One cell, a 2.5 x ATR stop with a 1 x ATR target, posted an out-of-sample t-statistic of 2.92 on 22 trades. That looks significant until you remember that 25 combinations were tested. At a 5% threshold, about one in twenty pure-noise tests will look significant by chance, so with 25 tries you should expect roughly one. Picking that cell afterwards would be curve fitting with extra steps.

Rules for refining a strategy

  • Change one decision at a time and give each change its own version number.
  • Write down why the change should help before you test it. A filter that improves results without a mechanism is noise you have selected.
  • Judge changes on in-sample data, then open the sealed window once. If the result is poor, the next step is a new idea, not a new parameter.
  • Prefer settings where neighboring values give similar results. A lone good cell surrounded by bad ones is luck.

How many trades you need before trusting a result

There is a short formula for how many trades it takes before a strategy's average result separates from luck. For a two-sided test at 5% significance with 80% power:

Trades needed = 7.849 x (standard deviation per trade)² ÷ (edge per trade)²

The 7.849 is the square of 1.96 plus 0.84, the standard normal values for those two settings. The ES pullback test (20 EMA pullback entries on 5-minute ES bars, August 28 to November 4, 2026) supplies the standard deviation: its trend-filtered version 2 had a net result per trade with a standard deviation of 8.25 points.

Edge per ES trade (points) Edge per ES contract Trades needed
0.5 $25 2,137
1 $50 535
2 $100 134
3 $150 60
5 $250 22

A real 2-point edge on ES, $100 per contract per trade, needs 134 trades before it separates from luck. A 1-point edge needs 535. The best version in that test, version 3 with added volatility and time filters, averaged +2.11 points with a standard deviation of 11.53, so it would need 235 trades. It had 17.

The same arithmetic explains why 30 trades is not a backtest. At a standard deviation of 8.25 points, the 95% confidence interval around the average after 30 trades is 2.95 points wide on either side. After 50 trades it is 2.29, after 100 it is 1.62, and after 200 it is 1.14. Most real edges are smaller than the 30-trade interval.

Why daily-bar strategies need years of trades

Slower timeframes do not make strategy validation easier. The same pullback rules (close back above the 20 EMA with the 20, 50 and 200 EMAs stacked, 1 x ATR stop, 2 x ATR target) on daily ES bars from November 2024 to November 4, 2026 (first 200 bars used for warm-up) produced 19 trades, averaging -18.81 points with a standard deviation of 120.69. The 95% interval ran from -73.08 to +35.46 points. A 20-point edge at that volatility would need 286 trades, which at this strategy's pace of roughly 16 trades a year is about 18 years of data.

While your count is short, trade the smallest size that keeps you in the game. Small size is how you keep collecting trades without letting one bad stretch end the experiment. Patience here is not passive. It is the part of the strategy that protects the sample.

Steps to build a trading strategy from idea to live size

Steps from idea to live size: write the idea in one sentence, seal part of your history, add one filter at a time, test the sealed window once and start small.

The order matters more than the speed. Doing these out of order produces rules that fit the past and fail in the next window.

Designing the strategy rules

  1. Write the idea as one sentence with no adjectives.
  2. Set process goals and pick a style and timeframe from the hours you actually have.
  3. Choose one market and confirm its contract size can express your stop at your risk budget.
  4. Turn the sentence into the six decisions: market, session, entry, stop, exit and size. Write down where any discretion lives.

Testing the strategy and going live

  1. Split your history by date before you look at it. Write the rules on the first part only and seal the rest.
  2. Count how many signals the raw idea produces before adding filters. A filter on a thin sample leaves you nothing to read.
  3. Add one filter at a time, with a written reason for each.
  4. Read win rate and payoff together and compare the result to the break-even win rate after cost.
  5. Open the sealed window once, run the frozen rules, and record the result without tuning.
  6. Compute the trades needed for your measured edge and volatility. If you are short, keep size small and keep collecting.
  7. Trade it live at minimum size, log every trade, and compare live results to the test.

Related lesson: Rule-based vs discretionary trading

Fitting a strategy to funded account rules

Funded and evaluation accounts add rules that a strategy must respect from the start. A typical firm's account has a daily loss limit, a trailing drawdown that ends the account if touched, a cap on contracts, and often a time by which positions must be closed. Those rules belong in the strategy spec, not in your head.

Three design changes follow. First, size from the drawdown, not the account balance: if the drawdown allowance is small relative to the balance, your real risk budget is the drawdown. Second, a strategy that holds overnight cannot run on an account that requires you to be flat each day, so the daily-bar version of a strategy is a different strategy there. Third, a daily loss limit should sit well outside your normal daily range of results, or ordinary variance will trip it.

On our accounts, every position has to be closed by 4:45 PM ET on regular trading days, with no overnight or weekend holds, so an intraday 5-minute strategy fits and a daily-bar swing strategy does not. Our Growth evaluations carry a daily loss limit ($1,250 on a 50K account), drawdown is an end-of-day trailing drawdown enforced in real time, and a 50K account can trade up to 4 minis or 40 micros. Our funded accounts are simulated.

How to Build a Trading Strategy FAQ

How do I build my own trading strategy?

Write your idea as rules covering the market, session, entry trigger, stop, exit and position size, with no judgement words. Split your price history by date, build the rules on the first part only, then test the frozen rules once on the sealed part. Compare the win rate with the break-even win rate after cost, and keep size small until you have enough trades for the average to mean something.

What is the 3-5-7 rule in trading?

The 3-5-7 rule is a risk guideline rather than a strategy. It usually means risking no more than 3% of the account on one trade and keeping combined risk across open trades under 5%. The 7 is used in different ways, either as a 7% profit target or as a reminder to plan winners larger than losers. It is not an official standard, and it still needs entry, stop and exit rules before it can trade.

How did one trader make $2.4 million in 28 minutes?

On March 27, 2015, an options trader bought out-of-the-money Altera options with a $36 strike for about $0.35 each, roughly $110,000 in total, one second after a news wire said Intel was in talks to buy Altera, according to Fortune. After a trading halt, the stock closed near $44.50 and the options were worth nearly $8.50, about $2.4 million more than 28 minutes earlier. Reuters later reported the trades came 19 seconds before a reporter's tweet, which points to a fast news-reading system rather than a repeatable strategy most traders could run.

Can I make $100 a day day trading?

Some days can end $100 up, but no strategy guarantees a daily figure, because the result depends on edge, size and trade count. On ES, $100 is 2 index points on one contract, and on MES it is 20 points. In the ES pullback test, one trade's result had a standard deviation of 8.25 points ($412.50 per ES contract), so treat $100 a day as an average to measure over many sessions, not a daily target.

Trading futures involves substantial risk of loss and is not suitable for everyone. Leverage means losses can exceed the amount you expected to risk, and fast markets can slip past stops. The test results in this guide describe past price behavior on specific instruments, timeframes and dates; they are illustrative, carry the limits of any backtest, and do not predict future results. Funded and evaluation accounts are simulated. Nothing here is financial advice or a recommendation to take any trade.