Quant tools
Three quant tools you'll reach for: estimate grid parameters, work out how much you'd save on fees, and run through how to set your API permissions. Everything is computed locally in your browser, and no data is uploaded — close the page and nothing's left behind.
Enter the range, grid count and capital to estimate how much one grid earns and what's left after both-side fees. Arithmetic grid, with a fixed price step per grid. Work it out before you build a bot on the exchange.
The step can't cover the fees. Widen the range or cut the grid count, or every filled grid loses money.
Net profit ≈ step as % of reference price − 2× one-side fee rate. It's a rough estimate of one grid filling once in the ideal case, and ignores slippage, unfilled orders and market direction. For reference only.
Grid trading slices a price range into a number of equal steps. Each step down, you buy a little; each step up, you sell a little — earning the spread step by step as price swings back and forth inside the range. In an arithmetic grid, the step per grid equals (upper bound − lower bound) divided by the grid count: the more grids, the smaller each step, so fills come more often but each grid's profit is thinner.
What really decides whether a grid earns is whether "the step as a percentage of the reference price" can cover "the both-side fee of buying plus selling." We use the range midpoint as the reference price, and subtract twice the one-side fee rate from the gross per-grid margin to get a rough net profit per grid. If that figure is near zero or negative, your grids are too dense and your range too narrow — the step still can't beat the fees, and the harder it runs the faster it loses. The tool flags this in red right away.
To be clear: this is a static estimate that assumes one grid is fully bought and then sold once. It doesn't account for whether orders actually fill, whether the market is one-way or choppy, slippage, or funding rates, and it doesn't predict how many grids will trigger over a period. The fee rate defaults to an example value; the real number, and the discount you get with a referral code, depend on what OKX currently publishes — check the OKX site and edit the field yourself. Treat it as a quick "do these parameters make sense" check, not a return forecast. For real parameters and a week of review, read our grid test article.
Based on your trade size and monthly trade count, estimate the fees you'd pay over a month and a year, and how much a referral discount saves you. Every figure is an example estimate, not a promise.
Example estimate, not a promise; the real fee rate, VIP tier and discount all depend on what OKX currently publishes. This calculates with the taker rate only — maker orders and unfilled orders aren't counted.
OK30001 get a fee discount.
Sign up for OKX with OK30001 →
Exchanges charge fees on spot and derivatives by trade size, usually split into maker (resting orders) and taker (filling orders) tiers, with taker typically the pricier one. For anyone running high-frequency quant or grids, fees are a fixed cost that happens every day and is easy to overlook — small on a single trade, but real once you multiply by a month's worth of trades.
The math here is plain: monthly volume equals trade size times trades per month; monthly fees before discount equal monthly volume times the taker rate; fees after discount equal the pre-discount fees times (1 − discount); the difference is what you save each month, times twelve for the year. It puts a number on whether "saving this much on fees" is worth caring about.
Worth stressing: the fee rate and discount on this page are example figures, not OKX's official quote and not our promise. Real rates shift with your VIP tier, holdings and promotions, and the discount a referral code brings also depends on what OKX currently publishes — check the OKX site and edit the input values yourself. Treat it as a reference for "is the referral code worth it, is building up volume to a better tier worth it," not a precise bill.
Before you create an API key on OKX and wire up a script, tick through these six. Ticks are counted locally in real time — nothing is saved, nothing is uploaded. This is the minimum bar for pushing quant funding risk below the floor.
Want to see exactly how and why to set each one? Read these two: OKX API risk control, hands-on · OKX API quant from scratch
Running quant over the API means handing part of your account's controls to a script and a pair of keys. Unlike a login password, an API key is made for a program to use, so it has more ways to leak — logs, screenshots, an accidental push to a public repo. That makes the security setup at key-creation time more worth your time than the strategy itself.
Of these six, the first two are the real floor. The first, "don't tick Withdraw": a quant script only needs to place orders and query, never to withdraw, so turn it off and a leaked key still can't move your money. The second, "IP allowlist": bind the key to the one IP your script runs on, and someone with the key still can't hit the endpoints. Skip either of these and the risk level is completely different, which is why the tool calls them out in red.
The other four narrow the risk further: store the key and Passphrase separately and keep them out of the code, get the flow working on the demo account before going live, isolate quant funds in a sub-account, and add per-trade and per-day caps as a hard brake. Ticking all six only means you've crossed the basic security floor — it doesn't mean absolute safety; but missing any one is a real exposure. This checklist doesn't collect your ticks — it's counted purely in your browser and resets to zero when you close the page.
The tools help you estimate parameters, estimate costs and run the checklist, but actually doing it still needs an account. New accounts that sign up for OKX with the referral code get a fee discount — it's the closest step to actually starting.
Crypto prices swing hard, and derivatives with leverage can wipe out your entire principal. Quant and bots don't guarantee profit, and the estimates in these tools are for reference only. Only use money you can afford to lose.