
Trading Hypothesis
You can explore the materials without registering, but with a free account you can track your progress and earn rewards along the way. By the end of your learning journey, you can have the knowledge you need to become a profitable Trader!
Join →Introduction
A trading strategy becomes truly useful when we do not simply say “we trade this pattern,” but can explain why we expect to have an edge in a particular market situation.
We call this idea a trading hypothesis.
A hypothesis is not a prediction and not a certainty. It is a testable assumption about what behavior we expect to recur, with what probability and under what conditions, in a defined market environment.
This way of thinking separates rule-based strategy development from simply searching for “good-looking charts.”
1. What Is a Trading Hypothesis?
A trading hypothesis is a statement that we can test using historical data and real trading results.
A simple example:
“If a stock in a strong uptrend pulls back to a previous support zone and then produces bullish confirmation, it has a higher probability of continuing the trend than a random entry point.”
This is a hypothesis because it:
• defines the market environment;
• defines the setup;
• defines the confirmation;
• states an expected behavior;
• and is testable.
The goal of a hypothesis is not to be correct every time. The goal is to find and measure a statistical edge over the long run.
2. Hypothesis ≠ Prediction
It is important to distinguish a hypothesis from a prediction.
A prediction might be:
“This stock will rise tomorrow.”
By itself, this is not a particularly useful strategic statement.
A hypothesis, in contrast, might be:
“In cases where the stock is in an uptrend on the higher timeframe, the pullback stops at a previous support zone, and bullish confirmation appears, trend continuation is more likely within the next X candles.”
This is:
• definable;
• measurable;
• testable;
• falsifiable.
That is the key distinction.
3. Why Do We Need a Hypothesis?
Without a hypothesis, we can easily end up explaining after the fact why a trade was good.
This is one form of hindsight bias.
For example:
“We bought because it looked very strong.”
That is not an adequate strategic rule.
Instead, we should be able to state:
• what trend was required;
• what setup was required;
• what confirmation was required;
• what entry occurred;
• what stop was used;
• what target was used;
• how much risk was accepted.
This allows the result to be evaluated objectively later.
4. Structure of a Hypothesis
A well-defined trading hypothesis usually contains several elements.
Market environment
Under what conditions is it valid? For example, trend, range, high volatility or low volatility.
Setup
What specific situation are we looking for? For example, pullback, breakout, retest or mean reversion.
Trigger
What activates the entry?
Expected behavior
What do we expect price to do?
Invalidation
When do we consider the hypothesis invalid for the trade?
Measurement horizon
Over what time horizon do we evaluate the result?
5. A Good Hypothesis Should Be Falsifiable
Falsifiable means that there must be an outcome capable of showing that our hypothesis does not work.
For example:
“Breakouts are good.”
That is not precise enough.
But:
“Daily breakouts that close above the previous 20-day range high, with above-average volume and ADX above 20, produce positive expectancy over the next 10 trading days.”
This is specifically testable.
If the data do not support it, we must reject or modify the hypothesis.
6. Hypothesis and Statistical Edge
The goal in trading is not for every trade to win.
The goal is to create a process with positive expectancy.
In simplified form:
Expectancy = (Win rate × Average win) − (Loss rate × Average loss)
For example:
• win rate: 50%;
• average win: +2R;
• loss rate: 50%;
• average loss: −1R.
Then:0.50 × 2R − 0.50 × 1R = +0.50R
This is positive expectancy.
A hypothesis therefore does not try to prove that we always win. It tests whether a rule set can produce positive expectancy under appropriate conditions.
7. Hypothesis and Setup
The setup tells us what situation we are looking for.
The hypothesis explains why we expect that situation to provide an edge.
For example:
Setup:
A pullback to previous support during a strong bullish trend.
Hypothesis:
If the bullish structure remains intact and previous support holds during the pullback, the probability of trend continuation may be higher.
The setup is therefore an observable market situation, while the hypothesis is the expectation associated with it.
8. Hypothesis and Entry
A hypothesis does not necessarily have to define every detail of the entry, but in a professional strategy these elements must be consistent.
For example:
Hypothesis:
After a pullback to support during a bullish trend, continuation is more likely.
Entry:
Break of the next candle’s high after bullish rejection.
Stop:
Below the structural swing low.
Exit:
Based on a predefined R-multiple or structural target.
This connects the hypothesis to actual execution.
9. Hypothesis and Invalidation
A good hypothesis must include a condition that shows when the idea has been invalidated.
Suppose:
“In a bullish trend, we expect trend continuation after the pullback.”
If price breaks the important swing low that was expected to preserve the bullish structure, the original hypothesis may be invalidated.
This does not necessarily mean that we should immediately open a trade in the opposite direction.
It simply means:the original assumption no longer holds.
10. Hypothesis and Timeframe
The same hypothesis can behave differently on different timeframes.
A setup may be successful on a daily chart while producing completely different results on a 5-minute chart.
Therefore, a hypothesis may need to specify:
• instrument;
• timeframe;
• trading session;
• holding period.
If these are not recorded, we can easily mix fundamentally different conditions.
11. Hypothesis and Market Regime
A strategic hypothesis does not necessarily work in every market environment.
For example, a trend-following hypothesis may work well:
• in a strong trend;
• with stronger momentum;
• with adequate liquidity.
The same strategy may perform poorly:
• in a sideways market;
• with weak momentum;
• when many false breakouts occur.
It is therefore important to define the market regime in which we expect an edge.
12. Hypothesis and Indicators
Indicators can help refine a hypothesis, but an indicator alone does not necessarily constitute a hypothesis.
For example:
“We buy below RSI 30.”
That is a simple rule, but not necessarily a good hypothesis.
A more complete hypothesis might be:
“In a range environment, after a downward overshoot at important support, RSI below 30 combined with bullish price action produces mean reversion more often than RSI alone.”
Here we have defined:
• the market environment;
• the location;
• the indicator;
• price action;
• the expected behavior.
13. Hypothesis and Volume
Volume can also be incorporated into a hypothesis.
For example, in a breakout strategy:
“A resistance breakout is more likely to continue when breakout volume is significantly above the average of the preceding period.”
This is testable.
We can examine:
• breakout volume;
• average volume;
• post-breakout return;
• false-breakout rate;
• holding period.
Volume thus becomes a measurable variable rather than simply something that “looks good.”
14. Hypothesis and ATR
ATR allows us to incorporate volatility into a hypothesis.
For example:“Breakout setup performance may differ between low-ATR and high-ATR environments.”
We can further classify:
• low ATR;
• medium ATR;
• high ATR.
We can then compare setup results.
The same strategy may be strong in one volatility regime and weak in another.
15. Test One Important Question at a Time
A common mistake is changing too many variables at the same time.
For example:
• new entry;
• new stop;
• new target;
• new timeframe;
• new indicator;
• new instrument.
If the result improves, we do not know what caused the improvement.
A better process can be:
1. one base hypothesis;
2. change one variable;
3. retest;
4. compare the result.
This brings us closer to identifying genuine cause-and-effect relationships.
16. Hypothesis and Overfitting
One danger of excessive optimization is overfitting.
This can occur when rules are adjusted so closely to historical data that the strategy looks excellent historically but performs poorly on new data.
For example:“We enter only when RSI is between 43.7 and 46.2, ATR is within an exact range, and volume exceeds its average by a specific amount.”
This may fit a historical pattern extremely well.
But it may have no real market rationale.
A good hypothesis is generally simple, logical and robust.
17. Hypothesis and Out-of-Sample Testing
A hypothesis developed on historical data should not be endlessly optimized on those same data.
We can use:
• in-sample data for development;
• out-of-sample data for validation.
If a hypothesis continues to perform acceptably on new data that were not used during development, this is stronger evidence that we did not simply fit it to the past.
18. Hypothesis and Journaling
A trading journal plays an important role in evaluating hypotheses.
For each trade, we can record:
• which setup occurred;
• the market environment;
• the timeframe;
• the confirmation;
• ATR;
• volume;
• entry;
• stop;
• result.
Later, these records can form a database.
The trading journal therefore becomes not only a record of the past, but also a strategy-development tool.
19. Successful Case Study
Suppose our hypothesis is:“In a strong bullish trend, a bullish rejection after a pullback to previous support is more likely to lead to trend continuation.”
We define:
• daily timeframe;
• liquid stocks;
• bullish market structure;
• at least one significant previous support;
• pullback;
• bullish rejection;
• structural stop;
• 2R target;
• maximum 10-day holding period.
We then test 100 historical cases.
Suppose:
• 54 winning trades;
• 46 losing trades;
• average win +2R;
• average loss −1R.
The hypothesis shows positive expectancy.
This does not prove that the strategy will work in the future, but it gives us reason to investigate it further.
20. Failed Case Study
Suppose a trader tests this hypothesis:“Every RSI reading below 30 is a reason to open a long trade.”
After 100 trades:
• 43 winners;
• 57 losers;
• many losses occurred during strong bearish trends;
• mean reversion was short-lived.
The hypothesis did not work adequately.
The trader does not necessarily conclude that RSI is useless.
The original hypothesis may simply have been too simplistic.
We could then test:RSI + range environment + important support + bullish confirmation
The modified hypothesis tests a different question.
21. A Hypothesis Is Not Sacred
A strategic hypothesis should change when the data justify it.
If testing shows that:
• it does not work in certain environments;
• it works better on another timeframe;
• one setup performs better;
• it is weak in certain market regimes;
we can refine the hypothesis.
The goal is not to prove that we are right at all costs.
The goal is to understand as precisely as possible when and why we may have a market edge.
22. Hypothesis → Test → Result → Refinement
One important strategy-development cycle is:
Hypothesis → Test → Data → Evaluation → Refinement → Retest
This is an iterative process.
The first hypothesis does not necessarily become the final one.
In fact, good strategy development naturally involves rejecting several hypotheses before finding a more robust approach.
23. What Makes a Good Trading Hypothesis?
A good hypothesis is:
• clear;
• specific;
• measurable;
• testable;
• falsifiable;
• based on a logical market environment;
• explicit about relevant variables;
• not dependent on hindsight explanations;
• not excessively optimized;
• verifiable on other data.
Most importantly:we define it in advance, not afterward.
24. Hypothesis Analysis Process
We can use the following process to develop our own trading hypothesis:
1. Define the market phenomenon we want to understand.
2. Define the market environment.
3. Define the setup.
4. Define the trigger condition.
5. State the expected outcome.
6. Define the invalidation conditions.
7. Define the timeframe and holding period.
8. Define how we will measure the result.
9. Test it on historical data.
10. Evaluate expectancy and other performance measures.
11. Modify the hypothesis if necessary.
12. Retest, preferably on separate out-of-sample data.
13. Only incorporate it into the strategy after sufficient evidence exists.
Practical Exercise
Formulate one of our own trading hypotheses.
Use this template:
Market environment:Under what conditions are we testing it?
Setup:What specific situation are we looking for?
Confirmation:What confirmation is required?
Entry:What activates the entry?
Invalidation:When do we consider the idea invalid?
Exit:What rule determines the exit?
Timeframe:Which timeframe are we using?
Holding period:How long do we hold the position?
Hypothesis:Exactly what market behavior do we expect?
Then find at least 30–50 historical examples and apply exactly the same rules to every case.
The goal is not to find the prettiest trades afterward, but to examine every relevant case consistently using predefined conditions.
Key Terms
Trading hypothesis: A testable assumption about the behavior we expect under defined market conditions.
Falsifiable: A statement for which objective data can show that it does not work.
Expectancy: The long-term expected result derived from the average outcomes of winning and losing trades.
In-sample: Data used to develop the hypothesis or strategy.
Out-of-sample: Data not used during development and reserved for validation.
Overfitting: Excessive fitting of a strategy to historical data, potentially reducing its ability to generalize.
Holding period: The time between opening and closing a position.
Invalidation: The condition at which the original trading idea is no longer considered valid.
Statistical edge: A repeatable advantage that can produce positive expectancy over a sufficiently large sample.
Knowledge Check
1. What is a trading hypothesis?
2. What is the difference between a hypothesis and a simple market prediction?
3. Why is it important for a hypothesis to be falsifiable?
4. How are a setup and a hypothesis related?
5. Why is it important to define the market environment?
6. What does positive expectancy mean?
7. Why is changing too many variables at once dangerous?
8. What is overfitting, and why is it a problem?
9. Why is out-of-sample testing important?
10. Why is the purpose of a hypothesis not to be correct on every trade?
Copying, reproducing, adapting, distributing, or publicly displaying any part or all of the Academy’s textual content without the author’s prior written consent is prohibited.
