Strategy review environment
Evaluate your AI trading bot before you trust its output.
A result is only as good as the assumptions behind it. Monoqant lays them out next to the numbers, checks them against your risk rules and keeps the record.
- Data
- Synthetic, seeded
- Execution
- Next-bar open
- Record
- Versioned + audit trail
Review · MA crossover 10/40
Simulation- Net result after costs — informationSample account, position 50% of equity9%hold: 6.79%
- Sample size — flaggedFewer than 10 round trips cannot separate skill from chance2min. 10
- Drawdown stop at 10% — within limitLargest peak-to-trough fall stayed inside the limit6.32%
- Single-bar loss limit 1.5% — flagged1 bar exceeded the limit-4.75%
- Costs as share of gross gains — within limitFlagged at 30% or more4.4%
Simplified simulation on synthetic data — not a full backtest. Change any control to re-run.
The point
A strategy result is a claim. Review the assumptions before the number.
Costs are assumptions
Fees and slippage are inputs you chose. Change them and the ranking of strategies can flip.
Regimes are assumptions
One market path is one possibility. A result that holds in only one regime is a description, not evidence.
Limits are assumptions
Risk rules only protect you if they are written down, measurable and checked every time.
Two scenarios, one strategy, every formula visible
Set scenario A and scenario B. The strategy stays fixed, so every difference comes from the market regime or the cost assumptions.
Scenario A
Scenario B
| Metric | A | B | Δ (B − A) |
|---|---|---|---|
| Net return(E_end − E_0) / E_0 | 18.72% | 17.73% | -0.99% |
| Gross resultΣ (exit − entry) · qty / E_0 | 18.89% | 18.77% | -0.12% |
| Total costsΣ fees + Σ slippage | 16.74 | 104.17 | 87.43 |
| Max drawdownmax_t (peak_t − E_t) / peak_t | 11.79% | 12.16% | 0.37% |
| Tradescount(round trips) | 2 | 2 | 0 |
| Time in marketbars in market / bars | 53.2% | 53.2% | 0% |
Sample data MA crossover 10/40, 100% position. Synthetic data with fixed seeds.
Six common ways to misread a strategy result
Each one is easy to make and easy to check — if the report shows the right numbers.
Reading gross as net
A headline return without fees and slippage describes a strategy nobody can actually run.
Trusting a small sample
Five trades can look like a pattern. Below ten, Monoqant flags the sample instead of praising it.
Testing one regime
A trend-follower in a trending market will look good. The question is what happens in the other three.
Ignoring the path
The same final number can hide a 30% drawdown. Ask whether you would have kept following the rules.
Tuning until it works
Trying many parameter sets and keeping the best one inflates results. Record how many variants you tried.
Comparing against nothing
Without buy-and-hold beside it, a positive number says little about the rule itself.
How a review is assembled
Six fixed stages, the same for every report — so two reports can be compared.
Parameters
Template, windows and position size, stored with the report.
Risk rules
Four measurable limits: position size, drawdown stop, single-bar loss, entry frequency.
Scenarios A / B
Two regime and cost settings for the same strategy.
Assumptions
Generated from the inputs, plus your own — listed, not hidden.
Results and checks
Every metric with its formula; every rule with pass or breach and the bar it happened.
Version and trail
Each save is a version; every change is logged with what changed.
Move a limit, see which events breach it
The sample strategy runs on the volatility-shock scenario. Every breach is listed with the bar it happened on and the measured value.
3 breaches · volatility shock
Simulation- bar lossbar change -1.51% beyond limit -1.5%
- bar lossbar change -9.16% beyond limit -1.5%
- dd stopdrawdown -10.4% reached limit -8% — flattened, trading halted
MA crossover 10/40 on synthetic data; limits apply only inside the simulation.
A record you can read later
What a report's audit trail looks like. Each line says what changed, from what, to what.
Report “Breakout review” · audit trail
Sample entries- v1 createdreport created · breakout 30 · A: steady trend · B: volatility shock
- checkB · drawdown stop 12% reached at bar 147 (−12.4%)
- v2 revisedrules.drawdownStopPct: 12 → 15
- checkB · drawdown stop 15%: within limit (worst −13.1%)
- v3 revisedb.costs.slippageBps: 5 → 15; assumptions: 0 → 1 items
- exportedmd
What is simulated, exactly
- Data
- 252 synthetic daily bars per scenario, generated from a documented model with a fixed seed.
- Execution
- Signal on the close, fill at the next open; slippage against you and a fee on every fill.
- Positions
- Long or flat; size as % of equity; no leverage, no shorting, no partial fills.
- Not modelled
- Order book, latency, market impact, financing, taxes, corporate actions.
Review notes
Designing Risk Rules That Can Be Audited
The Hidden Assumptions Inside a Backtest
How to Review an AI Trading Bot Without Chasing Headline Returns
Slippage, Fees, and the Gap Between Simulation and ExecutionShort answers
What does Monoqant do?
It reviews a strategy: it simulates it under two scenario/cost settings, checks the result against four risk rules, lists the assumptions and keeps a versioned report with an audit trail.
Who is it for?
Experienced individual users and small research teams who already work with strategy outputs and want a disciplined way to check them.
Do I need trading experience?
It helps. The terms are standard (drawdown, slippage, exposure) and every formula is documented, but the tool assumes you know why you are reviewing a strategy.
What do I get after confirming my email?
The Strategy Review workspace: up to 200 reports with versions, two-scenario comparison, four risk rules, audit trail and export as JSON, CSV or Markdown.
Is any real money or brokerage involved?
No. Monoqant does not connect to brokers or exchanges, does not hold funds and does not place orders.
Is this a backtest?
No — a simplified simulation on synthetic data with fixed costs and next-bar-open execution. It is designed to test reasoning and assumptions, not to estimate real performance.
What about AI?
Checks and statements are deterministic. If the operator enables an AI provider, a model can add a labelled prose explanation of a report. It can be wrong and does not replace the numbers.
What data is stored?
Your email, your reports with versions and audit trail, a session while you are signed in, and consent records. No passwords, payment details or IP addresses in our database.
How do I delete my data?
Settings → delete workspace data or the whole account; it takes effect immediately.
How do I unsubscribe from emails?
Newsletter emails have a one-click unsubscribe link; you can also switch the newsletter off in Settings. Sign-in emails are only sent when you request them.
Access
Start a strategy review
Confirm your email and the review environment opens. No password, no payment details.
- Unlimited versioned reports (up to 200)
- Two-scenario comparison with four risk rules
- Audit trail and export as JSON, CSV or Markdown