Scalping aims to capture small price changes over short holding periods, often with repeated entries and exits. That makes its first research question unusually plain: how large must the favorable move be before the trade can cover its actual round-trip costs? Define the instrument, order type, holding limit, size and decision interval before testing. “Scalp” by itself is not a testable rule.

Start with the fee floor

Fees are charged on executions and depend on the market, venue, account tier and schedule. At the 27 September 2026 source check, Hyperliquid’s published perpetual base tier listed a 0.045% taker rate and 0.015% maker rate per executed notional. Its schedule has other volume tiers and adjustments, so check the current fee table for the exact account and contract before using a rate.

For a fictional $10,000 position opened and closed at the same notional, two taker fills at that base rate cost $4.50 each: $10,000 × 0.00045 × 2 = $9, or 0.09% of notional. Two maker fills would cost $3 at the listed rate only if both executions qualified as maker. This arithmetic is a fee floor, not the total cost of the trade or a prediction that an order will fill.

illustrative scalping round trip cost floorAt the checked Hyperliquid perps base tier, two $10,000 notional taker executions cost $9 total. A separate assumed 0.04 percent combined adverse execution difference adds $4; the $13 sum is an illustrative break-even cost floor before funding or missed fills.$10,000 HYPOTHETICAL PERPETUAL · BASE-TIER TAKER ROUND TRIPEntry fee$4.50Exit fee$4.50Assumed fill friction$4.00Component bars use one common dollar scaleGROSS MOVE TO BREAK EVEN$9 + $4 = $13about 0.13% of $10,000Maker fees may be lower;a fill is not guaranteedFee schedule is venue-specific; assumed friction is fictional · funding and execution misses remain
At the checked Hyperliquid perps base tier, two $10,000 notional taker executions cost $9 total. A separate assumed 0.04 percent combined adverse execution difference adds $4; the $13 sum is an illustrative break-even cost floor before funding or missed fills. Illustrative teaching graphic · not live market data

Add execution friction, then test the target

Use one declared reference price for both sides of a round trip. If the combined adverse difference between that reference and the entry and exit executions is assumed to be 0.04%, it adds $4 on $10,000. Fees plus that fictional fill assumption total $13, so the gross favorable move must be about 0.13% just to reach zero before funding, latency, partial fills or other charges. The 0.04% input is deliberately an assumption; replace it with observations at the venue, order size and time horizon being studied.

Here is a quick sensitivity check using the same notional and the same 0.09% taker-fee floor. With no extra execution friction, the arithmetic floor is $9 (0.09%). At an assumed 0.04%, it is $13 (0.13%). At an assumed 0.10%, it is $19 (0.19%). These are scenarios, not measured spreads or estimates of typical slippage. If a fill model already executes at bid or ask and includes adverse price movement, do not add that same spread or slippage a second time.

That base tier is not a universal rate. Hyperliquid’s current page says fees depend on rolling 14-day weighted volume and shows separate schedules for perpetuals and spot. Its base spot taker rate is 0.070%, compared with 0.045% for perps; on the same fictional $10,000 round trip, that would be $14 versus $9 in taker fees before spread or slippage. Account volume, referrals, aligned quote assets and market-specific adjustments can change the effective rate. The article’s perp example should not be copied into a spot backtest.

As another check in the same simplified $10,000 notional model, a fictional 0.12% favorable gross move is $12. Under the middle cost assumption above, that is $1 below the $13 floor. A 0.20% move is $20, or $7 above that floor before omitted costs. Neither example says how often such moves occur or whether the signal can capture them. A strategy’s edge requires a complete sample of wins, losses, missed fills and costs, not a favorable example selected after the fact.

scalping break-even cost sensitivityFor a fictional $10,000 perpetual round trip at the checked 0.045 percent per-side taker base fee, the 0.09 percent fee floor is combined with three assumed adverse execution-friction levels. The resulting break-even gross moves are arithmetic scenarios, not observed costs or performance data.HOW THE BREAK-EVEN MOVE CHANGES WHEN FILL FRICTION CHANGESSame $10,000 notional and 0.09% round-trip fee floor · only the assumed extra execution cost varies0.00% extra0.09% · $90.04% extra0.13% · $130.10% extra0.19% · $190%0.10%0.19%Illustrative sensitivity only · not measured spread, slippage or a strategy result
For a fictional $10,000 perpetual round trip at the checked 0.045 percent per-side taker base fee, the 0.09 percent fee floor is combined with three assumed adverse execution-friction levels. The resulting break-even gross moves are arithmetic scenarios, not observed costs or performance data. Illustrative teaching graphic · not live market data

Repeated turnover raises the hurdle

The $9 fee floor is per completed $10,000 taker round trip at the published perp base tier. If a fictional process completes 20 such round trips a day, its fee arithmetic is $180 per day. Over 20 sessions that is $3,600, before any spread or adverse execution difference. This is not a forecast of scalping turnover or a return estimate; actual notional, account tier, maker/taker status and executions vary. It shows why counting each open-close cycle matters when the same strategy trades repeatedly.

Costs also change the break-even win rate. Suppose, only to isolate the arithmetic, every winning trade earns $20 gross and every losing trade loses $20 gross on that same notional. At $9 round-trip fees, the win rate needed to cover fees is 72.5%; with the assumed $4 execution friction added ($13 total), it is 82.5%; at $19 total costs it is 97.5%. Formula: (gross loss + round-trip cost) ÷ (gross win + gross loss). These are exact break-even thresholds for this symmetric toy distribution, not a recommended win rate, empirical hit rate or evidence that such trades are available. Real scalps have varied win/loss sizes, missed entries and market regimes; calculate expectancy from the full event ledger.

fictional scalping fee accumulation and break-even win-rate sensitivityLeft: at $9 in base-tier taker fees per $10,000 round trip, 1, 5 and 20 fictional round trips each day over 20 sessions cost $180, $900 and $3,600 before execution friction. Right: with equal $20 gross wins and losses, fee and friction totals of $9, $13 and $19 require break-even win rates of 72.5%, 82.5% and 97.5%. Scenarios are arithmetic, not market observations or expected performance.TURNOVER MULTIPLIES FEES · COSTS RAISE THE HIT-RATE HURDLE20 SESSIONS · $9 FEE PER $10,000 ROUND TRIPEQUAL ±$20 GROSS OUTCOMES · $10,000 NOTIONAL1 round trip/day$1805 round trips/day$90020 round trips/day$3,600Assumes same notional and base taker fee · no spread or slippagefees only · $972.5%fees + $4 assumed friction · $1382.5%fees + $10 assumed friction · $1997.5%0%50%100%Fictional arithmetic only · excludes other costs · break-even hurdle is not an observed win rate
Left: at $9 in base-tier taker fees per $10,000 round trip, 1, 5 and 20 fictional round trips each day over 20 sessions cost $180, $900 and $3,600 before execution friction. Right: with equal $20 gross wins and losses, fee and friction totals of $9, $13 and $19 require break-even win rates of 72.5%, 82.5% and 97.5%. Scenarios are arithmetic, not market observations or expected performance. Illustrative teaching graphic · not live market data

Maker pricing can trade fees for fill uncertainty

A resting limit order can avoid immediately crossing the spread, but its lower commission does not guarantee a lower realized cost. It may remain open, fill only in part, or execute before the price moves against the position. Hyperliquid describes its book as matching orders by price-time priority; that venue rule still does not let OHLC candles prove your queue position. Read the venue’s order-book description and model maker and taker execution as separate paths.

In the trade ledger, keep the signal time, order submission, acknowledgment, each partial fill, cancellation and final position close as distinct records. Include unfilled orders in the denominator when evaluating a signal: dropping them can make a limit-order test look better than the process that could actually have been followed. Stress the result with slower acknowledgments and a worse fill assumption rather than treating a candle touching the limit as proof of execution.

Funding belongs in the cost total only when the holding interval overlaps a funding settlement for that contract. Borrow, liquidation, network or builder charges depend on the instrument and venue. Keep each line item separate so that spread, slippage and fees do not silently overlap, and do not transfer a perpetual-contract cost model to spot trading.

Match the data to the claim

Hourly OHLC data cannot resolve second-level quotes, queue priority, order cancellations or network delay. Even one-minute bars may be too coarse when the rule depends on order-book changes. For those questions, use timestamped trades or order-book updates from the venue and preserve the feed, contract, fee tier, data gaps and assumptions with the result. A simulator that only has hourly synthetic candles can help compare slower rules or run a labeled cost sensitivity; it cannot establish live scalping performance.

scalping execution research audit workflowBefore judging a short-horizon strategy, define its rule and horizon, capture timestamped venue data, replay the order path including partial and unfilled orders, reconcile fees and funding, then stress execution assumptions and report every eligible signal.A SIGNAL IS ONLY THE FIRST RECORD IN AN EXECUTABLE TEST01 · Definesignal · hold limitsize · order type02 · Capturevenue · timestampstrades · order book03 · Replaysubmit · acknowledgepartial · cancel · fill04 · Reconcileeach execution feeeligible funding05 · Challengeno-fills · latencyhigher-cost casesKeep losing, rejected, partial and unfilled signals in the reportA candle touch cannot establish queue position, acknowledgment time or a completed fill
Before judging a short-horizon strategy, define its rule and horizon, capture timestamped venue data, replay the order path including partial and unfilled orders, reconcile fees and funding, then stress execution assumptions and report every eligible signal. Illustrative teaching graphic · not live market data

Before accepting a backtest, check whether the result survives a higher-cost scenario and a different market regime. Report rejected and unfilled signals alongside completed trades, and set a research stop condition before reviewing results. SuperQuant’s bundled examples use synthetic hourly data and simplified modeled costs; they do not validate rapid execution or demonstrate a scalping edge. Continue with the guides to trading styles, day trading and spread and slippage. This is educational material, not a trading recommendation.