1. Put every return on the same time basis
Do not compare a monthly figure with a yearly figure or a cumulative gain with an average monthly result. First establish what period each number represents. A strategy that gained 30% over two years is very different from one that claims 30% every month.
2. Compare net return, not just gross return
Performance fees, subscriptions, spreads, commissions, financing and slippage can materially change what the user actually keeps. If one service reports gross trading performance and another reports net account growth, the figures are not directly comparable.
3. Put drawdown next to return
Return without drawdown is incomplete. A strategy producing a high gain with a 50% peak-to-trough loss has a very different risk profile from one producing a lower gain with a 10% drawdown. Historical drawdown is not a future loss limit, but it is essential context.
4. Check how long the track record has existed
A strong three-month period can happen by chance or during favourable market conditions. A longer live record gives more information about how a system behaves across different environments. Duration should be considered together with trade count and consistency.
5. Look for consistency, not a perfectly smooth line
Real trading usually includes losing trades, weak weeks and negative months. An unusually smooth equity curve can be attractive, but it should also trigger questions about hidden risk, averaging, grid behaviour or whether the data is backtested rather than live.
6. Understand the capital structure
Leverage, funded-account structures, amplification programmes and ordinary cash accounts can make identical underlying strategy returns produce very different outcomes for the user's own capital. Always ask what capital base the reported percentage refers to.
Read: Amplification vs leverage →
7. Separate realised history from compounding projections
Compounding can make repeated monthly returns grow rapidly, but a projection assumes future returns continue and profits are reinvested. Treat compounding as scenario mathematics, not evidence of what will happen next.
Model a return and compounding scenario →
A simple comparison checklist
When comparing two bots, write down the same six fields for each: net return, measurement period, maximum drawdown, live-record duration, fee structure and capital/exposure model. If one of those is missing, the comparison is incomplete.
What matters most?
The best-looking return is not automatically the strongest proposition. A useful comparison asks how much return was produced, how much downside was required to produce it, how credible the evidence is and how much of the result the user actually keeps.