Spot Grid AI ranks USDT pairs on your connected venues by how well they suit grid trading — looking at volatility, range-boundedness, liquidity, volume, and stability. Each ranked pair comes with a calculated grid range, grid count, and capital allocation, validated against a 30-day backtest. One click deploys the parameters to a live grid bot.
Choose which connected exchange(s) to scan, pick LOW / MEDIUM / HIGH risk, set a minimum score, and optionally tell the scorer your capital.
Each USDT pair on the venue is scored across five dimensions. Pairs below threshold are rejected; the rest are ranked from EXCELLENT down to ACCEPTABLE.
For each surviving pair, the calculator derives upper and lower bounds, step size, grid count, capital per grid, and expected returns — tuned to the risk band you chose.
The proposed parameters run against the last 30 days of price data. The recommendation gets a VERIFIED tag if it cleared the backtest, or HIGH_RISK if it didn't.
Deploy recalculates the parameters one more time against fresh candles and spins up a live grid bot via the strategy engine. No copy-paste.
Each pair gets a 0–100 score that's a weighted blend of five dimensions, plus two hard rejects. The dashboard shows the per-dimension breakdown for every recommendation, so you can see why a pair scored where it did.
A weighted score can compensate weakness in one dimension with strength in another. Two checks override the score and outright reject the pair when triggered — because grid trading just doesn't work in those conditions.
A pair clearly trending below its long-term moving average and falling is refused. Grid trading buys dips and sells rallies — a market in steady decline keeps filling buys and never refills sells, and the bot drowns in inventory.
The grid's step size has to exceed twice the round-trip fee by a meaningful margin. Pairs with fees so high that the grid can't profitably cycle are rejected — no point recommending a setup the math kills.
Each pair shows its total score, the per-dimension breakdown, and a verbal label from EXCELLENT down to AVOID so you can sort by suitability.
Upper and lower bound, step size, number of grids, capital per grid, and expected daily and monthly return bands — all pre-computed for the risk band you chose.
A badge from the 30-day backtest: VERIFIED if the parameters cleared profitability, win-rate, drawdown, and trade-count checks. HIGH_RISK if they didn't — still surfaced, with the failure reason visible, so the decision stays with you.
Deploy recalculates the parameters one more time against the latest candles to avoid stale data, then creates a live grid bot through the strategy engine.
A background warmer scans the top USDT pairs per venue and refreshes the score cache on a half-hour cadence. Recommendations you see are within that window.
When you click deploy, the parameters are recomputed against the most recent candles. The cached version is the discovery aid; the live numbers are what your bot actually runs with.
The learning engine watches actual deployed-grid performance and adjusts volatility thresholds and prediction calibration. Combinations that consistently fail enter a cool-down so they're suppressed from new recommendations until they recover.
The analytics surface tracks every deployed recommendation against its expected return, so you can judge the system on outcomes — not on the marketing.
For each deployed grid, you see actual vs expected daily return, total P&L, cycles completed, and run date — so you can see which picks performed.
The current calibration multipliers and learned volatility thresholds are visible, so you can see how the learning engine has adapted to your account's actual results.
A few operational details worth knowing before you build a workflow on top of the ranked list.
The scorer covers USDT-quoted spot pairs. Perpetual futures, non-USDT quotes (USDC, BUSD, BTC pairs), and DEX-only tokens are separate surfaces.
Each pick is scored against the current market and expires after a day. To re-run a pick, request a fresh scan — the parameters get recomputed against the latest candles.
Every connected venue is scored independently against its own market. Cross-venue price comparisons — the same pair priced better elsewhere — are the Smart Order Router's job, not the scorer's.
You filter by venue, risk band, and minimum score. Sector or asset-class filters (meme coins, defi tokens, etc.) aren't in the UI today — the scorer weighs every USDT pair on the same five dimensions.
A pair that fails repeatedly on your account is suppressed from new recommendations for a period. That's protective — worth knowing so you're not surprised when a once-rejected pair takes a while to reappear even if conditions change.
If the score cache is cold (fresh server start), the first scan blocks while the warmer fills it. Subsequent requests hit the cache and return fast.
Scan, pick a ranked pair, click deploy. The grid bot does the rest.