# Kalshi Trading Bots and Polymarket Bots: The Fee Math Every Bot Has to Beat

> Kalshi trading bot or Polymarket bot? The break-even edge at 3 prices, why 100 one-contract orders cost more, and how copy trading erases a leader's 2¢ edge.

By Market Math Research · Updated 2026-11-17

A Kalshi trading bot or a Polymarket bot is a fee-paying machine first and a forecasting machine second. Every order it sends pays the venue's fee curve, and the bot only makes money if its edge per contract beats that fee, plus slippage, plus the cost of capital sitting in open positions until settlement. This is the math, not a coding tutorial. Every fee figure below renders live from our [fee registry](https://marketmath.io/fees) (verified November 16, 2026), so the numbers update when a venue changes its schedule.

The pipeline is the same for any bot: compute the fee at the price you trade, convert it to a break-even edge, decide maker or taker, size the position, then check what the API and settlement timing cost you on top.

**What a prediction market bot has to clear**

1. Fee at your price and order size
2. Break-even edge per contract
3. Maker or taker execution
4. Slippage and copy lag
5. Capital lock and rate limits
6. Size with Kelly

## Kalshi Trading Bot and Polymarket Trading Bot Break-Even Edge

The **break-even edge** is how many cents per contract your true probability must exceed the price just to cover fees. For a taker that holds to settlement on Kalshi or Polymarket, it equals the fee per contract, because both charge at the fill and nothing at settlement.

Kalshi's taker fee is roundup(0.07 × C × P × (1 − P)) per order. At 50¢ on a 100-contract order: 0.07 × 100 × 0.50 × 0.50 = $1.75, or 1.75¢ per contract. A bot buying YES at 50¢ needs a true probability above 51.75% before it earns a cent.

Here is the edge each venue demands from a taker bot, per contract, at three prices (100-contract orders):

| Venue and order type | At 20¢ | At 50¢ | At 80¢ |
|---|---|---|---|
| Kalshi taker | 1.12¢ | 1.75¢ | 1.12¢ |
| Polymarket taker, sports | 0.80¢ | 1.25¢ | 0.80¢ |
| Polymarket taker, politics | 0.64¢ | 1.00¢ | 0.64¢ |
| Polymarket US taker, Tier 1 | 1.52¢ | 2.38¢ | 1.52¢ |
| Novig taker, live | 0.96¢ | 1.50¢ | 0.96¢ |
| Novig, pre-game | 0.00¢ | 0.00¢ | 0.00¢ |
| ProphetX (fee on net winnings only) | 0.33¢ | 0.51¢ | 0.32¢ |

Read the cents as percentage points of probability. ProphetX is different: it takes 2% of net winnings per market at settlement and nothing on a loss, so the break-even probability is P / (P + (1 − P) × (1 − fee rate)). That formula produces the ProphetX row.

Polymarket US defaults to Tier 1 here (under $10K of 30-day taker volume). A bot that trades size moves down the schedule:

| Tier (30-day taker volume) | Θ | Fee on 100 contracts at 50¢ |
|---|---|---|
| Tier 1: Under $10K 30-day taker volume | 0.095 | $2.38 |
| Tier 2: $10K to $50K | 0.085 | $2.13 |
| Tier 3: $50K to $250K | 0.075 | $1.88 |
| Tier 4: $250K or more | 0.0695 | $1.74 |
| New account, days 1 to 7 | 0.0695 | $1.74 |

Two things in this table matter more than the headline rates.

**The fee is cheapest in cents at the extremes but most expensive per dollar deployed.** On Kalshi, the 20¢ fee is 1.12¢ on a 20¢ contract, which is 5.6% of the capital at risk. At 80¢ the same fee is 1.4% of capital. Longshot-buying bots pay four times the fee drag per dollar.

**Scalping bots pay twice.** A bot that enters and exits before settlement pays the taker fee on both fills. Buy 100 at 50¢ and sell at 52¢ on Kalshi: fees are $1.75 plus $1.75, or $3.50 against a $2.00 gross gain. Net: -$1.50. The same 2¢ scalp on a Polymarket sports market nets -$0.50. A 2¢ move is a losing trade for a taker bot on both venues. Model your own prices in the [fee calculator](https://marketmath.io/tools/prediction-market-fee-calculator) and the edge in the [PM EV calculator](https://marketmath.io/tools/pm-ev-calculator).

One more detail a bot has to code for: the fee is not uniform across markets. Each Kalshi series carries a `fee_type` and a `fee_multiplier` in the API (some MLB prop series run at half fees, a handful at zero). Each Polymarket market carries its own fee type, and older sports listings keep the rate they launched with. A bot that hardcodes one rate misprices its break-even on every exception. Read the per-market fields. The full schedules are in the [Kalshi fees guide](https://marketmath.io/blog/kalshi-fees-guide-2026) and [Polymarket fees explained](https://marketmath.io/blog/polymarket-fees-explained).

## Kalshi Bot Order Sizing: Why 100 One-Contract Orders Cost More

Kalshi rounds the fee up to the next cent **per order**, not per contract. Bots that slice orders into 1-lots to hide size or to chase queue position pay the round-up on every slice.

At 50¢, one contract costs 0.07 × 0.50 × 0.50 = 1.75¢, which rounds up to 2¢. So:

| Price | 100 orders of 1 contract | 1 order of 100 contracts | Extra cost of slicing |
|---|---|---|---|
| 50¢ | $2.00 | $1.75 | 14% |
| 20¢ | $2.00 | $1.12 | 79% |
| 5¢ | $1.00 | $0.34 | 194% |

The round-up does the most damage far from 50¢, where the true fee is a fraction of a cent and every 1-lot still pays a full cent. A 5¢ longshot bot sending 1-lots pays 2.9 times the fee of a bot sending one order. The fix is mechanical: batch child orders into the largest size the book can absorb without moving price. Kalshi API batch calls do not help on fees, because each order in a batch is still its own order.

## Kalshi Market Maker and Polymarket Market Maker Bots: Maker vs Taker Fees

A **maker** order rests on the book and gets filled by someone else. A **taker** order crosses the spread and fills immediately. Most of the fee math for bots comes down to which side you are on.

| Venue | Taker bot pays | Maker bot pays | Maker extras |
|---|---|---|---|
| Kalshi, most series by count (non-sports and minor series) | roundup(0.07 × C × P × (1 − P)) | $0 | None |
| Kalshi, maker-fee series (every major sports series: NFL, NBA, NHL, college, soccer games, spreads, totals, championship futures) | roundup(0.07 × C × P × (1 − P)) | roundup(0.0175 × C × P × (1 − P)) | None |
| Polymarket | C × category rate × p × (1 − p) | $0 | Maker rebate: a share of taker fees, paid daily (the program changes Nov 20, 2026; see [prediction market rebates](https://marketmath.io/blog/prediction-market-rebates)) |
| Polymarket US | Θ × C × p × (1 − p), Θ by tier | $0 | Rebates only at very high monthly maker volume |
| Novig | Coefficient × C × P × (1 − P) on live and futures | $0 | Maker Credit eligibility on live trades |
| ProphetX | 2% of net winnings per market | 2% of net winnings per market | Monthly make-volume rebate above $50K of make volume |

Kalshi's maker fees apply only to series flagged with maker fees, roughly 160 of more than 16,000 series. That is few by count, but it includes every major sports series: NFL, NBA, NHL, college football and basketball, soccer games, spreads, totals and championship futures. Makers there pay 0.0175 × P × (1 − P) per contract. On those series a 100-contract maker fill at 50¢ costs $0.44, against $1.75 for the taker. The 1-lot round-up hits makers too: 100 maker 1-lots at 50¢ cost $1.00 versus $0.44 as one order.

ProphetX is the outlier. Makers and takers pay the same fee on winnings, so the maker advantage there is the spread you capture and the volume rebate, not a lower rate.

The maker side is where small bots have a structural edge, and also where they take **adverse selection**: the fills you get are disproportionately the ones informed traders wanted. A zero fee does not make a quote profitable if the price moves through it after news. Spread capture, inventory limits and adverse selection are covered in depth in [prediction market market making](https://marketmath.io/blog/prediction-market-liquidity). The short version for a bot: a resting quote is worth the spread minus the expected loss on toxic fills, and on Polymarket plus any maker rebate you qualify for.

## Polymarket Copy Trade Math: How a Leader's 2¢ Edge Disappears

A **copy trading bot** watches a leader's trades and mirrors them. On Polymarket this is easy to build because every wallet's trades settle on Polygon and are public. The math is harder than the code.

Assumptions, stated so you can change them:

- The leader buys YES at 50¢ with a true probability of 52%. That is a 2¢ edge.
- The leader rests a limit order and fills as a maker, paying $0 on Polymarket and on Kalshi series without maker fees. On every major Kalshi sports series (NFL, NBA, NHL, college, soccer games, spreads, totals, championship futures) the maker pays 0.0175 × P × (1 − P), so the leader's 100-contract fill at 50¢ costs $0.44.
- The copier sees the fill, then crosses the spread as a taker. By then the leader's buying has lifted the price 1¢ (illustrative slippage). The copier pays 51¢ plus the taker fee.
- 100 contracts. Edge per contract = 52¢ − copier's price − fee.

| Venue | Leader edge (maker) | Copier fee per contract at 51¢ | Copier edge per contract | Copier EV on 100 contracts |
|---|---|---|---|---|
| Kalshi, series without maker fees | 2.00¢ | 1.75¢ | -0.75¢ | -$0.75 |
| Kalshi, major sports series | 1.56¢ | 1.75¢ | -0.75¢ | -$0.75 |
| Polymarket, sports | 2.00¢ | 1.25¢ | -0.25¢ | -$0.25 |
| Polymarket, politics | 2.00¢ | 1.00¢ | 0.00¢ | $0.00 |
| Polymarket US, Tier 1 | 2.00¢ | 2.37¢ | -1.37¢ | -$1.37 |

The leader keeps 2¢ on Polymarket and on Kalshi series without maker fees, and 1.56¢ on a major Kalshi sports series after the maker fee. The copier keeps a fraction of the leader's edge at best and loses money on most venues. One cent goes to slippage and most of the rest goes to the fee. The best case in the table is a low-fee Polymarket category such as politics, where the copier keeps 0.00¢ per contract, well under half of the leader's 2¢. Copying a taker leader is worse still: on Kalshi a taker leader nets 0.25¢, and the copier one step behind is deeply negative.

The break-even rule for Polymarket copy trading or Kalshi copy trading: **the leader's edge must exceed your slippage plus your taker fee at the price you fill.** At 50¢ on Kalshi with 1¢ of lag, that is more than 2.75¢ of edge per contract. Few traders sustain that, and the ones with the biggest public profits are often makers whose edge is the spread itself, which a taker copier cannot copy by definition.

Two more problems sit outside the table. Leaderboards show survivors, so a top wallet's past return overstates its expected return. And a leader's position is sized for the leader's bankroll and correlation book, not yours. Size each copied trade on your own estimate with the [Kelly criterion calculator](https://marketmath.io/tools/kelly-criterion), which returns zero when the edge after fees is zero.

## Arbitrage Bots: Two Legs, Two Fees

Cross-venue arbitrage bots pay a fee on each leg, so the fee hurdle stacks. Buying both sides at 50¢ as a taker on Kalshi and on a Polymarket sports market costs 3.00¢ per contract pair before resolution-rule risk. The full worked math, including partial fills and settlement mismatches, is in [prediction market arbitrage after fees](https://marketmath.io/blog/prediction-market-arbitrage-after-fees).

## Polymarket API Rate Limits, Locked Capital and Withdrawal Time

Fees are the visible cost. Three other costs decide whether a bot strategy that works on paper survives live.

**API rate limits.** Both venues meter automated traffic, and the designs differ in ways that matter to a bot:

- **Kalshi** uses token buckets. Each request costs tokens, your access tier sets the refill rate, and reads and writes (orders, amends, cancels) draw from separate buckets. Higher tiers are earned through API use and trailing trading volume. Over the limit, requests return HTTP 429 and the bot must back off. Batch requests do not save budget, since each item is billed separately.
- **Polymarket** limits are IP-based and enforced by Cloudflare on sliding windows, per endpoint, with burst and sustained limits on order placement and cancellation, plus per-signer limits on trading. Over the limit, requests are throttled (delayed or queued) rather than rejected.

The cost is not the limit itself. It is what happens to a quoting bot when its cancel is delayed. On Kalshi a rejected cancel leaves a stale quote live until the retry. On Polymarket the cancel arrives late with no error at all. Either way the stale quote gets picked off by exactly the informed flow described above. A market-making bot that cancels and replaces on every tick burns write budget fastest. Check the current tier tables in each venue's docs before designing the quote loop.

**Capital locked until settlement.** A contract ties up its full price until the market resolves or you sell. Say a maker bot nets 1¢ per contract after fees on 50¢ contracts, a 2% return per cycle. If the market settles in 3 days and the capital recycles, that is about 2% × 365 / 3 = 243% annualized before variance. If the market settles in 120 days, it is about 2% × 365 / 120 = 6.1%. Same edge, a 40x difference in return on capital. A bot should rank opportunities by edge per day of lockup, not edge per trade. The [prediction market liquidity guide](https://marketmath.io/blog/prediction-market-liquidity) covers how depth limits how much capital you can deploy at that edge.

**Withdrawal time.** Kalshi and other US exchanges pay out over bank rails. Polymarket pays out in USDC on Polygon, which still has to be converted to dollars elsewhere, with its own fees and timing. A vendor's return figure should be net of both.

## Checklist Before You Pay for a Polymarket Bot or Copy Trading Service

No endorsements here. Run any Kalshi bot, Polymarket copy trading bot or signal service through these questions before paying:

1. **Are the returns net of current fees?** Ask whether the backtest uses each market's actual fee type and the order type the bot really sends. A backtest that assumes maker fills while the live bot takes is fiction.
2. **What is the maker and taker split?** Ask for the share of fills that were maker. A copy service is taker by construction.
3. **What is the measured lag?** For copy trading, ask for the average price difference between the leader's fill and the copier's fill. Compare it to the leader's edge using the table above.
4. **How are orders sized on Kalshi?** Bots that slice into 1-lots pay the per-order round-up on every slice.
5. **Is the track record verifiable?** Polymarket wallet history is public onchain. Screenshots are not a track record. Ask how many strategies were run and dropped.
6. **What does the service charge on top?** Add subscription fees and any per-trade fee the front end adds to your break-even edge.
7. **Who controls the money?** A service that asks for a wallet private key or seed phrase can move your funds. API keys should be scoped to trading with no withdrawal access wherever the venue allows it.
8. **What happens at a rate limit?** Ask how the bot handles HTTP 429 on Kalshi and throttling on Polymarket, and whether quotes are pulled when the connection degrades.
9. **How long is capital locked, and where?** Ask for average days to settlement, and confirm the venue is available to you. Polymarket's international exchange and Polymarket US are different venues with different fees and eligibility.

If a seller cannot answer the first three with numbers, the edge is unmeasured. Run the claimed edge, your fee and your lag through the [PM EV calculator](https://marketmath.io/tools/pm-ev-calculator) yourself.

## Sources

- [Kalshi fee schedule (PDF)](https://kalshi.com/docs/kalshi-fee-schedule.pdf)
- [Kalshi API docs: Rate Limits and Tiers](https://docs.kalshi.com/getting_started/rate_limits)
- [Kalshi API: series endpoint with fee_type and fee_multiplier](https://api.elections.kalshi.com/trade-api/v2/series)
- [Polymarket docs: Fees](https://docs.polymarket.com/trading/fees)
- [Polymarket docs: Rate Limits](https://docs.polymarket.com/quickstart/introduction/rate-limits)
- [Polymarket docs: Builder Program overview](https://docs.polymarket.com/developers/builders/builder-intro)
- [Polymarket US fees](https://docs.polymarket.us/fees)
- [Novig: Fees on Novig](https://support.novig.com/en/articles/16195057-fees-on-novig)
- [ProphetX trading fee schedule and make-volume rebate (CFTC filing)](https://www.cftc.gov/filings/orgrules/rules08152617271.pdf)
- Fee inputs: [Market Math fee registry](https://marketmath.io/fees), verified November 16, 2026

## Frequently asked questions

**Does a Kalshi trading bot need a bigger edge than a Polymarket bot?**

As a taker, usually yes. Both venues use a fee curve of rate times price times one minus price, but Kalshi rounds up to the cent per order and its standard rate is higher than most Polymarket categories. As a maker, Polymarket charges nothing, and Kalshi charges nothing on most series by count, but every major Kalshi sports series (NFL, NBA, NHL, college, soccer games, spreads, totals, championship futures) charges makers a fee on the same price times one minus price curve, at the maker rate in the table above. The live break-even table on this page shows the current numbers.

**Does Polymarket copy trading work?**

Only when the leader's edge per contract is larger than your slippage plus your taker fee. A copier fills after the leader, at a worse price, and pays the fee the leader often avoided by resting a limit order. A 2 cent edge with 1 cent of lag is typically gone after fees.

**Why do small orders cost more on Kalshi?**

Kalshi rounds the fee up to the next cent on each order. One contract near 50 cents carries a fraction of a cent above one cent in raw fee and rounds to two cents, so a hundred 1-contract orders cost more than one 100-contract order. Far from 50 cents the gap widens.

**What are the Polymarket API rate limits?**

Polymarket sets limits per endpoint and by IP, enforced by Cloudflare on sliding windows, with burst and sustained limits on order placement and cancellation. Excess requests are delayed rather than rejected. Check the Rate Limits page in Polymarket's docs for the current figures.

**Should a Kalshi market maker bot worry about maker fees?**

Yes, on any sports bot. Series flagged with maker fees are a small share of all series by count, but they include every major sports series: NFL, NBA, NHL, college, soccer games, spreads, totals and championship futures. Makers there pay the Kalshi maker rate (shown in the maker vs taker table above) times price times one minus price, rounded up per order. Every other series charges resting orders nothing. Read each series' fee type from the API before quoting.

---

Source: Market Math (Market Math Research), fees verified November 16, 2026. https://marketmath.io/blog/prediction-market-trading-bots-fees-math
