Why you should know this
A backtest can make almost any idea look convincing if rules are adjusted after seeing the data; the skill is designing a test that the future could still prove wrong.
Academy 12 is where ideas become operating rules. The aim is not to collect strategy names. It is to learn how to ask the same professional questions of every style: what is the decision rule, what market behavior is it trying to exploit, what assumptions support it, what costs sit between the signal and the result, and what evidence says the method no longer fits?
How the strategy actually makes a decision
Backtesting applies fixed strategy rules to historical data to estimate how the method would have behaved. The dangerous part is that the researcher already knows the past. Rules can be unconsciously adjusted until they fit the historical sample, creating an impressive result that never existed before the researcher saw it.
A trustworthy process separates rule design, in-sample exploration and out-of-sample evaluation. It also includes realistic costs, signal timing and asset availability. If a strategy uses today’s list of successful tokens to test the past, it may contain survivorship bias before the first trade is calculated.
What must be true before the strategy makes sense
The test needs an exact asset universe, timeframe, signal definition, entry timing, exit timing, position sizing, fees, spread/slippage assumptions and data-cleaning method. Look-ahead bias must be prevented: the strategy may use only information that would have been available at the simulated decision time.
Sample size matters, but more trades are not automatically more independent evidence. Hundreds of signals generated during one bull market may still represent one dominant regime.
Work through the decision, not just the definition
Suppose a backtest shows a 70% win rate before costs. After adding 0.25% total friction per round trip, the average win falls from 1.0R to 0.75R while the average loss stays near 1.0R. The headline win rate did not change, but expectancy changed materially.
Now suppose the rule was tuned using ten different lookback values and only the best result was reported. That is another hidden cost: selection bias. The proper test should show whether nearby parameter values also produce sensible results.
Where the strategy gives its edge back
A strategy can have a sensible market idea and still produce a poor result if its costs, timing or operating conditions are wrong. Before judging performance, separate market edge from execution drag. Fees, spread, slippage, missed signals, funding or borrowing costs, tax-record obligations and unavailable liquidity may matter differently for each style. The next lesson in this family turns those frictions into an explicit testing ledger rather than leaving them as footnotes.
Backtests fail through look-ahead bias, survivorship bias, unrealistic fills, ignored delistings, missing fees, overfitting and repeated testing on the same “out-of-sample” period. They also fail when the trader treats historical profitability as proof that the market mechanism will persist.
A good backtest produces questions for forward testing. It does not end the research process.
Skill practice — build the rule before seeing the answer
Write one complete strategy rule set before opening the historical outcome. Split the data into development and untouched evaluation periods. Record trade count, win rate, average win, average loss, expectancy, drawdown, turnover and cost assumptions. Then run a parameter-sensitivity check and remove the best trade. If the strategy survives only one narrow setting or one exceptional winner, label it fragile rather than “proven.”
Do not score the exercise only by whether the hypothetical trade made money. Score whether the rule was clear enough that another reader could make the same decision from the same information. A lucky outcome from an undefined process is not the skill Academy 12 is trying to build.
How this connects to market mastery
A strategy becomes useful only when it can be compared with alternatives, tested under different regimes and retired when its assumptions fail. That is the bridge from “I know what this strategy is called” to “I can decide whether this strategy belongs in this market and in my operating constraints.”
*Cryptocurrency and virtual asset transactions are highly volatile and irreversible, may result in significant losses, and do not guarantee returns; customers should trade only after understanding the risks involved.
