Compare against buy-and-hold, honestly
You have a strategy curve and the control from unit 1. Putting them on one chart is easy; comparing them fairly is not.
Same window, same costs, same data
Three ways a comparison quietly cheats, all of them accidental:
Different windows. The strategy starts at the first bar where its slow average exists — bar 50, if the slow window is 50. The control started at bar 1. Those are different periods, and if bars 1 to 50 fell, you just handed the strategy a free head start. Start both at the same bar.
Costs on one side. The strategy trades and pays; buy and hold trades once. If costs are off for the control they should be off for both, and they should be off for neither.
Different columns. Both must use adjusted close. Mixing them makes the comparison meaningless in a way that produces a plausible-looking chart.
Time in the market is not the same as risk
A strategy that is flat half the time carries about half the market exposure, so a lower return is expected rather than a failure — and a higher return with half the exposure is a genuinely interesting result, not just a bigger number.
So print time in market as a percentage beside the return. Without it, "beat the benchmark by 3%" is uninterpretable: it might be skill, or it might be leverage in disguise.
The comparison that actually means something
Not "did it beat buy and hold" — over a single window, with one rule, that question is nearly noise. The useful questions:
- Did it beat the control after costs, in the window where both were live?
- What did it give up to do it — how much time out of the market, how many trades, how deep a drawdown?
- Is the gap larger than the gap you would get by changing one parameter slightly? If a 20/50 crossover wins and 21/50 loses, you found noise.
That last one is a test you can run in a minute, and it is the cheapest reality check in this course.
Two curves, one chart, four numbers
Plot both lines from the same start. Under them, four numbers per line: total return, time in market, number of trades, max drawdown. Nothing else yet — unit 4 adds the rest, and a wall of statistics this early just gives you more places to find something flattering.
Try it now
Run your strategy at its chosen parameters and at four neighbours — fast ±2, slow ±5. Put the five results side by side. If the chosen one is a spike among losers rather than the middle of a plateau, you have learned something important before unit 3 even starts.