The backtester that tells you when your edge is fake.
AIM = Automatic Investment Management — rules that buy more as prices fall and trim as they rise.
Runs your strategy across five rules-based engines, benchmarks it against buy-and-hold, and validates every result out-of-sample.Volatility & risk research, without the hype.
0
rules-based AIM engines
0-way
strategy comparison
Walk-fwd
validation, every optimize
~0s
to your first verdict
Real screen: Sharpe 2.59 in-sample collapses to -0.65 out-of-sample — the Lab calls it overfit.
From ticker to verdict in ~30 seconds
Most backtesters cost you an evening of configuration. AIM Lab is one step: type a symbol, hit Enter, and get a regime-aware verdict — engines compared, risk gates checked, out-of-sample validated.
Type a ticker
SPY, TSLA, BTC — anything on daily bars.
Hit Enter
Engines run, regime is read, gates check.
Read the verdict
A full, honest answer in about 30 seconds.
Real screens. Real verdicts. No idealized dashboard.
Everything below is the actual AIM Lab interface — the same views you get on day one.
Five AIM engines, one honest scoreboard
Run Classic AIM, Volatility-Enhanced, VA-AIM, Zero-Cost Average and Constant Value against DCA and an auto-optimized config — seven columns, one winner, every metric that matters.
- CAGR, Sharpe, max drawdown & trade count
- Transaction costs and ending allocation
- A hard “vs Buy & Hold” column on every strategy
Ranked #1 doesn’t mean deploy-ready
Every comparison gets a plain-English verdict tied to the current market regime. The Lab will happily tell you a strategy topped the table — then flag that it failed five risk gates and is research-only.
- Regime read: trend, efficiency, bars-in-regime
- Risk gates that block unsafe “winners”
- “Why this verdict?” you can actually inspect
Every equity curve, every trade, and why it fired
Overlay all engines on one chart, then drill into a single strategy’s entries, buys and sells. Prices are valued on adjusted close, so results line up with reality — not a cherry-picked window.
- Equity overlay across all engines
- Buy/sell markers you can click to see the trigger
- Adjusted-close pricing (splits & dividends)
We’ll tell you when doing nothing wins
When DCA or plain buy-and-hold beats every engine, the Lab says so. When an optimized config looks curve-fit, it warns you. Plain-English cards — explicitly not financial advice.
- Flags overfit configs (train vs unseen gap)
- Surfaces “no meaningful difference” honestly
- Tells you when passive beats active
Real run: the Lab credited the winner, then flagged the optimized config as likely overfit — in the same breath.
We tell you when NOT to trade
Every optimizer run is re-scored on data the search never saw — and the Walk-Forward panel is stamped with the exact config it validated. When every candidate fails out-of-sample, you don’t get a leaderboard of curve-fit configs dressed up as winners. You get one verdict: keep your baseline.
- “No deployable edge found” — said in plain red letters
- Validation is stamped with the exact config it graded
- One-click “Keep baseline” — the honest default
Real run: all 8 candidates failed validation — in-sample +42.94% became −35.43% on unseen data. One verdict: keep your baseline.
Trade the regime you’re actually in
Markets don’t have one personality — they switch between them. AIM Lab reads the current regime and matches it to the strategies proven to work there, instead of the average of all of them.
Live regime detection
Every dataset and position is classified in real time — strong uptrend, downtrend, choppy or moderate — with transition alerts when a read starts to wobble.
Recommended for today
The moment a regime is detected, the strategies proven to shine in it are one click away. Diagnosis to strategy, instantly.
Per-regime, per-ticker analytics
See exactly how any preset performed in each regime on each instrument — because TSLA in chop is not SPY in chop.
Four questions, four verdicts — not an indicator wall
Most trading tools pile on indicators until every chart argues with itself — RSI says overbought, MACD says momentum, the moving average says hold. No signal outranks another, so you become the tiebreaker, on every trade, at decision speed. AIM Volatility Lab is built the opposite way: a conflict that arrives resolved is information; a conflict that arrives unresolved is noise.
Is the edge real?
01The out-of-sample validation verdict
Every optimized configuration is re-tested on data it never saw. Sharpe, CAGR and heatmaps are the evidence behind that verdict — not five more opinions competing with it.
Does it fit the market now?
02The regime read — with one trust state
One label, one qualifier — solid, transitioning, or too young to trust — and its age. If the ensemble is split, the verdict stamps itself “provisional.”
How much?
03Percent of account at risk
Position sizing runs through your risk rules before a campaign opens — the check either passes or it blocks, with the dollar amount at stake stated plainly.
Am I following my plan?
04Expectancy per trade + your discipline score
Dollars earned per trade taken, and how many trades in a row you’ve closed without breaking a rule. The metrics that matter to a human, not a stats appendix.
And when the data can’t support a confident answer, the verdict says so out loud — “verdict provisional” — because a tool that hedges honestly beats one that’s confidently wrong.
This week's duels — the win and the loss, both published
Every week the Lab races its five engines against buy-and-hold and publishes the widest win AND the widest loss. Most tools show you one of those.
The QR on each card links back to the live run — same five mechanical engines, same capital, shared end-of-day bars. Hypothetical simulation, not advice; past results don’t predict future ones.
Two expectancies, one execution gap
A backtester can only grade the strategy. The built-in trade log grades the other half of every result: the person executing it. Both run the same expectancy math — average outcome per trade taken — but they answer two very different questions.
Backtest expectancy
Does the SYSTEM have an edge?
Thousands of simulated trades, validated out-of-sample, net of costs and slippage. This is the number the whole lab exists to defend — and it says nothing about you.
Trade-log expectancy
Do YOU execute it?
Every fill you actually log — your prices, your fees, your exits — scored as expectancy per trade, in dollars and in R. The same math, pointed at the only variable the backtester can’t simulate.
When the first is positive and the second is negative, that difference is the execution gap — your behavior, not your strategy — and the journal tags usually name the culprit.
Journal tags name the leak
Tag trades as you log them — revenge, FOMO, early exit, plan — and the dashboard automatically names your worst-performing tag. The culprit gets a label, not a vibe.
Sample-size honesty
Under ~20 closed trades, your expectancy is marked provisional — a six-trade streak is a coin-flip dressed as a statistic, and the journal refuses to pretend otherwise.
Discipline, counted
Every stop, target and breach is recorded against the plan you set at entry. Your discipline score counts rule-clean closes in a row — measured, not remembered.
Takes under a minute a trade. Free on every tier, no limits — and your logged history exports any time. It’s yours.
Everything you need to test an edge — and kill a bad one
Eight modules, one workflow: backtest, compare, optimize, validate — in seconds, not an evening. Built for people who’d rather find out now than in the market.
Backtest
Type a ticker, hit Enter — a full verdict in about 30 seconds. Any AIM engine, adjusted-close pricing, transaction costs and slippage baked in.
Compare
Seven strategies side by side with a single winner column, each benchmarked against plain buy-and-hold.
Optimize
Parameter search that re-runs on the train window and scores its winner on an untouched holdout — then warns you if it looks overfit.
Session Diff
Pit two saved runs head-to-head inside Compare — did your change actually help? Keep the receipts: shareable, reproducible, exportable to PDF.
Regime detection
Reads trend, efficiency and choppiness so every verdict fits the market you’re actually in.
Risk Manager
Guardrails and gates that block unsafe “winners” before they reach your allocation.
Allocator
Turn a validated strategy into position sizes and deployable cash — not just a backtest.
Community
What worked — and what didn’t — for traders like you: validated edges, shared strategies, receipts one click deep.
What AIM actually is
AIM — Automatic Investment Management — is a rules-based method that mechanically buys more as prices fall and trims as they rise. No forecasting, no discretion, no vibes.
AIM Volatility Lab implements five variants — Classic, Volatility-Enhanced, VA-AIM (Value Averaging-Enhanced), Zero-Cost Average and Constant Value — then stress-tests each against buy-and-hold and out-of-sample data, so you can see where the rules help and where they don’t.Tap any variant for a plain-English definition and example.
“A comparison table that separately answers ‘who ranked best,’ ‘on what fair basis,’ ‘with whose money,’ ‘within whose risk rules,’ and ‘does the edge survive validation’ — without ever crowning a lie.”
AIM Volatility Lab is in open beta, built by a small Bao Research team that values validation over hype. No inflated user counts, no fabricated testimonials — just the tool and its verdicts. See how we grade ourselves →
What you can actually test
Assets
US equities & ETFs, plus crypto pairs (e.g. BTC, ETH)
Resolution
Daily OHLC bars
Pricing
Adjusted close — splits & dividends adjusted; may differ from a brokerage’s raw prices
Data
Market data via providers such as EODData and Yahoo Finance
Your AI agent is a great interface. It's a terrible calculator.
Ask a chatbot to backtest and it will invent numbers that merely look right. Connect it to AIM Lab instead — the agent does the talking, our verified engines do the math.
Native MCP server, one URL. Add AIM Lab as a custom connector in Claude, ChatGPT or any MCP-capable agent — no glue code, authenticated with your own scoped API key.
Deterministic, audited answers. Every result comes from the same regression-tested engines and adjusted market data the app uses — not from a one-off script the agent improvised and nobody reviewed.
- Real tools, real portfolio. Backtests, regime reads, verified presets and your live positions — exposed as agent tools:
run_backtestget_regime_readlist_presetslist_positions
Pick a plan that fits how much you test
Start free. Upgrade when the signal proves itself. Cancel anytime, self-serve.
Free
The basics, forever free.
free forever
No card required
- 5 market tickers / day
- 3 optimize · 3 Compare runs / day
- 1 Matrix run / day (3 core engines)
- 1 account · 2 portfolios (1 monitored)
- 10-ticker watchlist, weekly re-rank
- Current-regime banner on datasets
- Daily alert checks · Morning Brief headline
- Contributes anonymized strategy results to Community — the collective intelligence every member sees
Pro
For active, multi-strategy traders.
or $790/yr — save 17%
- The five-engine race with out-of-sample walk-forward validation — know when an edge is fake
- No daily ticker limit — ad-hoc backtest any symbol
- 30 optimize · 15 Compare · 3 Matrix runs / day
- 3 accounts · 20 portfolios (8 monitored)
- 25-ticker watchlist, re-ranked nightly · hourly alert checks
- Regime Intelligence — pattern context, transition alerts & weekly digest
- Recommended for today — regime-matched presets
- Full Morning Brief — spike cautions, flips & actions
- Share strategies to Community · collective-data opt-out
- Everything in Free
Premium
Depth and automation for the systematic trader.
or $1490/yr — save 17%
- API access — keys for scripts, agents & LLM/MCP integrations
- Priority compute — front of the queue, always
- Per-regime, per-ticker analytics — TSLA in chop is not SPY in chop
- 150 optimize · 50 Compare · 9 Matrix runs / day
- 100-ticker watchlist · 25-ticker race batches, nightly priority re-rank
- 6 accounts · 60 portfolios (20 monitored) · 15-minute alert checks
- Everything in Pro
One payment. Premium for life. 150 seats, ever.
- Everything in Premium — forever, no renewal
- No daily ticker limit — backtest any symbol
- Founder badge on Community strategy shares
- Premium fair-use caps (currently 150 optimize runs / day)
- API access — keys for scripts, agents & LLM/MCP
- Locked in before public-launch pricing
The fine print, in writing: lifetime means the Premium feature set as it exists today, for the life of the product, with the same fair-use compute caps as Premium, as those caps stand from time to time (currently 150 optimize runs/day) — founders always get whatever Premium gets. Future add-on products may be sold separately. One seat per person, non-transferable. Once all 150 seats are claimed, lifetime seats will never be offered again.
50 seats at this price
30-day money-back guarantee — full refund, and your seat returns to the pool.
Research tool, not advice. Backtested and simulated results are hypothetical, do not reflect live trading, and do not guarantee future performance. AIM Lab is an educational tool and is not financial, investment, or trading advice.
Security, before you trust us with anything
A research tool for serious traders should be boring about safety. We are.
Two-factor auth
TOTP apps, email codes and backup codes — so a leaked password isn’t enough.
Device management
See every session, revoke any device, and get new-login email alerts.
Hash-chained audit exports
SHA-256 chained records and HMAC-signed certificates auditors can verify.
Your data, your call
One-click GDPR export and full account erasure, no support ticket required.
Secure checkout powered by Stripe. Cancel anytime — self-serve, right from your account — and you keep access until the end of the period.
Questions, answered
The honest version. Still curious? Reach out to the Bao Research team.
Find out if your edge is real — before the market does.
Backtest, compare, optimize and validate in one honest workflow. Free to start — open beta, no invite needed.