Futures backtesting software for CME data Backtests that fill
like live trading.

Atlantive backtests CME futures on one-second bars since 2010 and on every order of the book. Your limit orders wait their turn behind the real orders ahead, after your latency, and every result is net of fees.

No card needed.

ESU6Sell 1 at 7,480.00

fast-forward 2:56:08.035 a.m.ET

29 contracts ahead, in 18 orders

0 traded 0 canceled

A touch fill assumes your order was first in line. This one had 29 contracts ahead of it, and the exchange filled it 3 minutes later. Atlantive fills from the same line: on 706,925 resting ES orders of July 10, 2026, it matched the exchange 98.1% of the time.

The most accurate fills in futures backtesting. Measured, not promised.

of 10,146,414 real CME orders filled at the exchange's own instant and size
98.54%
of resting ES orders the exchange filled on July 10, 2026, matched by Atlantive, against touch fills
98.1%vs 44.4%
one-second bars since June 6, 2010, on every CME Globex futures contract from its listing
2010
a full run over 1,980,426 one-second bars
3.51 s

Most accurate: of the platforms compared on this page, the only fills checked against the exchange's own, as of October 20, 2026.

Backtest vs live Five places
backtests lie.

Most backtests drift from live trading in the same five places, and mostly in the trader's favor. Atlantive replaces each assumption with the exchange's own record, then measures what the assumption would have cost.

Where backtests drift from live trading, and what Atlantive does instead
WhereThe usual assumptionWhat Atlantive doesMeasured
01Limit ordersFilled the moment price touches themFilled in their place in line, from every order of the book 98.1% vs 44.4% of resting ES orders the exchange filled, matched by Atlantive and by touch fills
02StopsFilled at the stop priceFilled against the book that was really there, after your latency if your platform holds the stop 5.7% of 1,233 NQ stops held at CME in one overnight session filled at their price; the median paid 3 ticks
03Inside the barStop or target picked from open, high, low and closeRead from one-second bars or the book 43 of 85 bars in one overnight ES session where the guess picked the wrong exit
04CostsSlippage and fees left at zero, or one flat numberEvery fee per contract, slippage by depth, break-even slippage on every result Every result prints its fees, its slippage and the slippage it breaks even at
05Rollovers and sessionsA stitched chart, roll gaps booked as P&L, trades on holidaysThe real dated contract on CME's own calendar Every session and every holiday from the exchange calendar

Limit order fills and queue position A limit order
waits its turn.

On CME, a resting limit order fills only after the orders ahead of it at its price. Atlantive puts yours behind every real order that was there first. It fills when the line ahead has traded or canceled, as the exchange would have filled it.

Exact place in line, from every order of the book

Atlantive replays CME's market-by-order record: every order, its size and its priority.

Your order joins at the back of its price, and the line ahead shrinks only by what really traded or was canceled. No trade at your price, no fill. On every resting ES order the exchange filled on July 10, 2026, 706,925 of them, Atlantive matched the exchange's own fill 98.1% of the time. A touch fill matched 44.4%, and one in a hundred was off by 160 seconds or more.

Partial fills, liquidity used once

A 10-lot fills in the pieces the market allowed, at the moments it allowed them.

What your order takes is gone for the rest of the run, so two of your orders never fill against the same contract. A real 10-lot sell of ES at 7,477.75 filled in five pieces, 1, 2, 2, 4 and 1, inside 446 microseconds.

Latency before every order and cancel

Every order, change and cancel reaches the exchange after your latency: 10 milliseconds by default, from 0.1 milliseconds to 10 seconds, with a random delay you can repeat exactly.

In ten real milliseconds the line ahead of one ES order grew from 2 contracts to 16. A cancel that arrives after the fill it was meant to stop is a fill, and the run books it.

Size that walks the book

An order that can fill right away takes the best price first, then the next, as deep as its size needs, against the book as it stood when the order arrived.

Size beyond what the book showed fills by a square-root market-impact model, its constants on the method page, never at a price nobody offered.

Order types and trade management Every order type.
Every change mid-trade.

Atlantive takes every order a futures trader uses and every change made while a trade is open. Each change follows CME's priority rules: an order you move loses its place, an order you shrink keeps it, exactly as on Globex.

  1. 3:05:00.010 Buy 3 at 7,476.75, a limit at the bid Joined, 9 ahead
  2. 3:05:32.649 Filled 3 at 7,476.75, 32.6 s later; 8 ahead canceled first
  3. 3:05:32.659 Target: sell 3 at 7,479.75 Joined, 20 aheadStop: sell 3 at 7,474.75, held at CME
  4. 3:05:39.611 Another trader's order ahead of the target is changed Back of the line, 18 ahead, then 17
  5. 3:06:09.093 Scale out: sold 1 at 7,478.25 at market Target cut from 3 to 2 Kept its place, 16 ahead, then 16Stop to break-even, 7,476.75 for 2 Trigger changed, back of the line
  6. 3:07:28.128 Stop triggered by a trade at 7,476.75: sold 2 at 7,476.75

Changes mid-trade, and your place in line

Move the stop to break-even, scale out a third, trail the rest, add on a pullback.

Cut an order's size and it keeps its place. Change its price, its type or its trigger, or raise its size, and it goes to the back of the line, as CME's rules require. Every change is stamped on the trade, with the line ahead before and after.

Every order type

Market, limit, stop, stop-limit, market-if-touched and trailing stops.

Brackets and one-cancels-other groups that stay in sync through partial fills. Iceberg orders that show part of their size. TWAP, VWAP and percent-of-volume orders that work a position over time. Day, GTC, GTD, IOC (fill and kill, in CME's words) and FOK.

One-contract stops in the overnight session of July 28 to 29, 2026: how they filled against the real book
StopEvery stop by its slippageAt the stop priceMedianWorst
ES Held at CME , 1,109 stops66.5%0 ticks15 ticks
ES Your platform, 10 ms , 1,109 stops47.4%1 tick20 ticks
NQ Held at CME , 1,233 stops5.7%3 ticks42 ticks
NQ Your platform, 10 ms , 1,233 stops4.9%5 ticks40 ticks

Stops slip on the real book

A stop triggers when a trade prints at or through its price.

A stop held at CME then fills against the book at once, within the exchange's protection band. A stop your platform holds pays your latency first. Either way it fills against the bids or offers that were really there, so a fast market costs what it cost. In the overnight session of July 28 to 29, 2026, when the book is thinner than by day, a fixed rule placed a stop 4 ticks either side of the price at the start of every minute. A backtest that fills stops at their price books every one of those ticks as profit.

Tick data, Level 2 and order book backtesting Every depth, from one-second
bars to every order.

Pick the depth a strategy needs. One-second bars cover every CME Globex futures contract since June 6, 2010. The book adds the best bid and offer, ten levels a side, or every single order, and each depth fills the way its data allows.

The four depths of CME data Atlantive backtests on
DepthWhat it holdsHow orders fillBest for
Open, high, low, close and volume of every secondFills At the next one-second close after the decision, with slippage per sideBest for Bar strategies over years of history
The best bid and offer with their sizes, and every tradeFills Market orders at the best bid or offer, up to the size shown; resting limits behind the size shown at their price; after latencyBest for Market orders and the cost of the spread
Size and order count at ten prices a sideFills Walks the levels; place in line estimated from the size aheadBest for Size, depth signals, imbalance
Every order, its ID, its size and its place in lineFills Exact place in line, partial fills, latencyBest for Limit orders, scalping, anything that rests
ESU6, 2:56:08.035 a.m. ETMarket by order (Level 3)

The hero's order, 19th of 19 at 7,480.00. Contracts / orders at each price.

ESU6, July 28, 2026, 8:20 p.m. ETA long from the open, target and stop 8 ticks away

The five-minute bar

The same five minutes, 300 one-second bars

248 seconds traded

43 of 85 five-minute bars of that night where the guess picked the wrong exit

Inside the bar, no guessing

A five-minute bar shows where price went, not in what order.

Platforms guess the path from open, high, low and close. Some let you switch on a finer series: TradingView's and MultiCharts' Bar Magnifier, NinjaTrader's Order Fill Resolution, TradeStation's Look-Inside-Bar. Atlantive reads the path from one-second bars or the book.

Take a real ES bar of July 28, 2026, at 8:20 p.m. ET, and a long from the open with a target and a stop 8 ticks away. The guess says the stop. The seconds say the target, hit 20 seconds in. In that overnight session, the guess picked the wrong exit on 43 of the 85 bars where such a bracket fit.

Filled after the decision, never on it

A decision made on a bar's close fills at the next one-second close, never at the close that decided it.

Decided at 3:02:00 a.m. ET on July 29, 2026, at 7,477.00. Filled at the next one-second close, 3:02:01 a.m., at 7,477.00 plus the default tick of slippage: 7,477.25.

One-second bars since
June 6, 2010
One-second bars served
3.7B+
  • Equity index ES NQ YM RTY
  • Interest rates ZN ZB ZF SR3
  • Energy CL NG RB HO
  • Metals GC SI HG PL
  • Currencies 6E 6J 6B 6A
  • Agriculture ZC ZS ZW LE
  • Micros MES MNQ MGC MCL
  • And every other CME Globex future

Coverage

Every CME Globex futures contract: equity index, interest rates, energy, metals, currencies, agriculture and micros.

Over 3.7 billion one-second bars, every one served from Atlantive's servers: nothing to download, nothing to clean.

Futures slippage and commissions Every cost, until
the edge breaks.

A futures edge is often a tick or two a trade. Atlantive charges every fee per contract and every tick of slippage, then tells you how much more slippage a strategy could pay before it stops making money.

1 tick
$2.10

The run's net as slippage rises

One ES contract, 26 fills

The run's result, cost by cost
Before costs 13 round trips at their prices+$1,437.50
Commissions 26 fills at $2.10 a side−$54.60
Slippage 26 fills at 1 tick, $12.50 a tick−$325.00

Net result

+$1,057.90

4.26 ticks a side of slippage, and the run breaks even Printed on every result

A demonstration run. Your own runs print the same figures. The opening range breakout of the landing's simulated ES market, 13 round turns.

  • Break-even slippage on every result

    Every bar result prints its break-even slippage: the net with no slippage, divided by the tick value and the contracts filled. If a strategy breaks even at half a tick per side, it has no edge you can trade.

  • Fees as charged on the day

    Exchange and NFA fees per contract from the schedule in force on each trade date, plus your broker's clearing fee and commission. ES, NQ, CL and the micros each pay their own.

  • Slippage by depth

    On bars, slippage is a setting. By default it is what each market was measured to charge: one tick per side on ES, two to three on NQ. Set it from zero to 20 ticks, or as a share of the bar's range. On the book it is not a setting. It is what the book charged.

  • An execution report on every run

    Fill rate and time in line. Slippage against the price that triggered each signal. The fills that went straight against you, and where price went in the seconds after each fill. Cancels that arrived too late. Everything a desk measures, in words a trader uses.

Sessions, rollovers and exchange rules The exchange's calendar,
contract by contract.

A backtest that trades a holiday, holds through a halt or books a roll gap as profit is testing a market that never existed. Atlantive trades the real dated contract on CME's own calendar, under the exchange's rules as they stood that day.

Rollovers the way traders roll

Orders always go to a real contract, ESU6 rather than a stitched series.

The roll happens on the day the market moved to the next contract, or a set number of days before expiry, and it is booked as a trade with its own costs. Back-adjusted series are there for signals, never for fills.

Sessions, holidays and early closes

Each product trades its own hours, from the Sunday 6 p.m. ET open to the Friday close, with the daily break, holidays and early closes from the exchange calendar.

An order placed when the market is shut waits for the open.

  • Halts, price limits and reopening auctions

    Price limits, circuit-breaker halts and stop-logic pauses apply as CME applied them. While a market is halted your orders wait; when it reopens, they fill at the reopening price, as far as the auction's traded size allows. Crude oil's negative prices of April 20, 2020 replay as they printed.

  • Margin, settlement and account rules

    Intraday and overnight margin, the exchange's settlement prices, position limits, and forced liquidation at your broker's margin cutoff, with presets to start from. A portfolio runs as one account, and prop firm rules come from the strategy's own trades on the prop page.

  • Spreads, micros and options on futures

    Calendar and inter-commodity spreads trade as instruments of their own. Micros sit beside their full-size contracts. Options on futures run in the same backtest as the futures they hedge.

Python futures backtesting engine Python in,
native code out.

Write a strategy in Python, or describe it to the agent. Atlantive compiles it to native code before the run. One engine thread replays 2.28 million order-book events a second.

strategy.py Python
1 from atlantive import BUY, NEW_YORK, SELL, Bars, RollingMinMax, Session, Strategy
2
3
4 class OpeningRangeBreakout(Strategy):
5 # One-second bars, filled at the next price a tick against us.
6 data = "bars"
7
8 # Params are the tunable knobs. A sweep sets them by name, with no edit.
9 class Params:
10 stop_ticks: int = 24
11 target_ticks: int = 48
12 min_range_ticks: int = 24
13 clip: int = 1
14
15 class State:
16 clock: Session(zone=NEW_YORK)
17 bars: Bars(interval_ns=300_000_000_000)
18 # 09:30 to 09:45 New York: the high and the low of its three bars.
19 opening: RollingMinMax(window=6)
20 stop: float
21 target: float
22 traded_day: int
  1. Checked against the strategy language
  2. Compiled to native code
  3. Replayed on the exchange's record

Speed, with its method

order-book events a second, on one engine thread
2.28 M
a whole run of 1,980,426 one-second bars, 871 to 1,707 ns of engine time a bar
3.51 s
a sweep by default, 4 at once, each a full backtest
48 runs

2.28 million order-book events a second on one engine thread: ESU6, 428,921 events of July 29, 2026, on an AMD EPYC-Milan, medians of seven paired rounds. A whole run over 1,980,426 one-second bars takes 3.51 seconds, the container's start included.

  • Write it, or describe it

    Strategies are short Python files in a closed trading vocabulary, checked before they run. Prefer plain English? The AI agent writes the file, runs it and shows you every trade.

  • Sweeps and simulations at scale

    Run a parameter grid of 48 settings by default, four at a time, each a full backtest. Rerun a strategy under thousands of latency and slippage draws to see its spread of outcomes, not one lucky path. Walk-forward and the rest of the validation protocol run with the agent.

  • Same run, same result

    Same data, same code, same settings: the same orders, fills and trades, every time. Each result carries its receipt: data version, engine build, contract map, session calendar, fill model, latency and slippage. Across two engine builds, 62 of 62 runs came out identical.

  • Take it to your platform

    Take a finished strategy where you trade: export it to NinjaTrader NinjaScript to run it automated, or to TradingView Pine Script to chart it and send alerts.

Backtest accuracy Checked against the
exchange's own fills.

We took 10,146,414 real resting orders out of the exchange's record and sent each through the engine as if it were ours. 98.54% filled at the exchange's own instant and size.

ES and NQ, seven sessions, July 8 to August 4, 2026

98.54%

of 10,146,414 real resting orders filled at the exchange's own instant and size

at the same instant as a slow, plain rebuild of CME's matching rules
99.94%
never filled by the engine, 0.43%
43,631
Orders matched to the exchange's own fills, by session and contract
SessionOrders95%Matched100%Share matched
Jul 8 ES1,107,41098.37%
Jul 8 NQ481,50399.28%
Jul 10 ES706,92598.10%
Jul 10 NQ363,66699.22%
Jul 13 ES872,97598.10%
Jul 13 NQ439,83099.15%
Jul 14 ES792,69698.24%
Jul 14 NQ406,29099.23%
Jul 17 ES1,156,12498.33%
Jul 17 NQ532,88099.22%
Jul 23 ES1,154,91798.46%
Jul 23 NQ576,13299.23%
Aug 4 ES1,010,83997.84%
Aug 4 NQ544,22799.04%

The axis starts at 95%, so the sessions can be told apart. Engine as frozen on September 14, 2026.

The check: how every order was checked

Every resting order the exchange filled on seven sessions of ES and NQ: July 8, 10, 13, 14, 17 and 23, and August 4, 2026. A match means the same instant and the same size as the exchange's fill. ES matched between 97.84% and 98.46% a day, NQ between 99.04% and 99.28%. Against a slow, plain rebuild of CME's matching rules, 99.94% filled at the same instant. The engine never filled 43,631 of these orders (0.43%). Orders canceled unfilled are outside this check. Engine as frozen on September 14, 2026.

Tested like infrastructure

cases asserted against a slow, plain reference, across sixteen families of order streams
88,959,214
strategy runs reproduced step for step, about 1,210,000 steps
456
automated tests in every release gate
11,796
runs identical across two engine builds
62of 62

What no backtest can know

Five things are in no exchange record. Here is how each is handled.

  1. 01Size others hid.

    Iceberg reserves and stops waiting to trigger, at the exchange or a broker, never appear in the book. Atlantive fills against what was shown, and hidden size counts only once it trades.

  2. 02How the market would have answered you.

    A replay does not react. Small orders fill exactly; size beyond the book fills by the square-root impact model.

  3. 03Your own latency.

    It depends on your connection and your broker. Pick the profile that matches your setup.

  4. 04Spread liquidity implied from other contracts.

    CME publishes it by price, not by order, so it has no place in line.

  5. 05Your broker's rules.

    Margin cutoffs and liquidation policies differ by firm. Set yours, or start from presets.

Compare backtesting software How each platform
fills a limit order.

Every backtester assumes something about fills. Here is what each one assumes, from its own documentation, as of October 2026. Of these, only Atlantive fills by place in line on CME data and publishes its match with the exchange.

How each backtesting platform fills a limit order, from its own documentation, verified October 20, 2026
Resting limit ordersInside the barSlippagePublished check against real fills
AtlantiveFilled from their place in line, on every order of the bookRead from one-second bars or the bookWhat the book charged; on bars, set per side98.54% of 10,146,414 real orders
TradingViewBy default, “guaranteed to fill” once price reaches them; optional ticks of penetrationPath guessed from open, high, low, close; Bar Magnifier on Premium plansA constant number of ticksNone
NinjaTraderFilled from bar prices; place in line not describedEach bar split into three virtual bars, or a finer seriesA fixed number of ticks, on market, stop-market and MIT ordersNone
TradeStationFilled from a finer interval with Look-Inside-Bar; place in line not describedLook-Inside-Bar, from tick to dailyA fixed amount a trade or a contractNone
MultiChartsFilled when price touches, or exceeds by a set number of pointsBar Magnifier, from ticks or minutesA settingNone
Sierra ChartQueue position tracked only on charts that are not replayingA bar-based test reads the bar's open, high, low, closeNot describedNone
QuantConnectFilled on price alone, “completely and immediately”The latest tick, quote bar or trade barZero by defaultNone

Default backtest behavior as each vendor documents it, verified October 20, 2026. Settings can change some of it. Corrections welcome at support@atlantive.ai.

Specifications The engine,
in one table.

Every default and every limit of the engine, in one place. Each one is a setting you can see on every result.

Engine

Engine
Native code, event by event, one thread a run
Strategies
Python, compiled to native code before the run
Clock
Exchange time to the nanosecond; no wall clock inside a run
Repeatability
Same data, code and settings: the same orders, fills and trades
Throughput
2.28 million order-book events a second on one thread; 871 to 1,707 ns of engine time a one-second bar
Sweeps
A grid of 48 settings by default, four at a time
Isolation
No network, read-only data

Market

Data
One-second bars on every CME Globex futures contract since June 6, 2010; the order book at every depth
Instruments
Futures, micros, calendar and inter-commodity spreads, options on futures
Sessions
Each product's own hours, holidays and early closes
Exchange events
Price limits, halts, stop-logic pauses, reopening auctions
Rolls
On the day the market rolled, or N days before expiry; booked as a trade
Continuous series
Back-adjusted or ratio-adjusted, for signals only; fills on the real contract
Account
Portfolios; intraday and overnight margin; settlement prices; forced liquidation at your margin cutoff
Export
TradingView Pine Script, NinjaTrader NinjaScript

Execution

Depths
One-second bars, top of book, ten levels, every order (market by order)
Queue
Price-time priority, behind every real order there first; no fill without a trade
Partial fills
Each trade fills what reaches the order; the rest keeps its place; liquidity used once
Order types
Market, limit, stop, stop-limit, market-if-touched, trailing stop, iceberg, TWAP, VWAP, percent of volume; OCO groups and brackets
Time in force
Day, GTC, GTD, IOC, FOK
Changes
Size cut keeps its place; price, type, trigger or size up goes to the back
Latency
10 ms by default; presets of 0.1, 2, 10, 30 and 120 ms; up to 10 s with a repeatable random delay; a cancel never overtakes its order
Bar fills
At the next one-second close, never the close that decided it
Bar slippage
Each market's measured slippage by default (1 tick per side on ES, 2 to 3 on NQ); 0 to 20 ticks, or a share of the bar's range
Market impact
The book walked level by level; size beyond it by a square-root impact model
Fees
Exchange and NFA fees per contract from the schedule on each trade date, plus your clearing fee and commission

Questions about futures backtesting.

How accurate is futures backtesting?

As accurate as its fill model. Atlantive replayed 10,146,414 real ES and NQ resting orders through its engine as if they were its own, and 98.54% filled at the exchange's own instant and size. Touch fills, the default elsewhere, matched 44.4% of resting ES orders. Orders fill in their place in line, after your latency and every fee. What no backtest can know, such as hidden iceberg size, is listed beside the method.

Can I backtest futures for free?

Yes. Sign up to Atlantive free, with no card, and run backtests on CME futures data straight away, from one-second bars since 2010 to every order of the book, net of every fee and tick of slippage.

How far back can I backtest futures?

One-second bars for every CME Globex futures contract go back to June 6, 2010, served by Atlantive, so there is nothing to download. The order book replays at every depth, from the best bid and offer to every single order.

Do limit orders fill when price touches them?

Not on the exchange, and not in Atlantive. A resting limit fills only after the orders ahead of it at its price have traded or canceled. On every resting ES order the exchange filled on July 10, 2026, a touch fill matched the exchange 44.4% of the time; Atlantive matched 98.1%.

What is queue position in a backtest?

Queue position is how many contracts sit ahead of your order at its price. Atlantive reads it from CME's market-by-order record, where every order has an ID and a priority, so your place is exact and the line drains only by real trades and cancels. On Level 2 data it is estimated from the size ahead.

How much slippage should I expect on ES and NQ?

It depends on the order, the contract and the hour, which is why Atlantive takes it from the book. In the overnight session of July 28 to 29, 2026, one-contract stops held at CME filled at their price 66.5% of the time on ES, with a median of 0 ticks; on NQ only 5.7% did, with a median of 3 ticks. Held by a platform 10 milliseconds away, the medians were 1 and 5 ticks.

How do stop orders fill in a fast market?

In Atlantive a stop triggers once a trade prints at or through its price, then fills against the bids or offers really in the book, level by level: at once if CME holds it, after your latency if your platform does. In the overnight session of July 28 to 29, 2026, the worst ES stop paid 20 ticks and the worst NQ stop 42.

What happens when the stop and the target are inside the same bar?

Atlantive reads which came first from one-second bars or the book, instead of guessing from the bar's open, high, low and close. In the overnight session of July 28 to 29, 2026, that guess picked the wrong exit on 43 of 85 five-minute ES bars where an 8-tick bracket fit.

Can I backtest order flow and Level 2 data?

Yes. An Atlantive strategy can read and trade on the best bid and offer, ten levels a side, or every single order. Each depth fills the way its data allows, and market-by-order fills use your exact place in line.

What is L1, L2 and L3 market data?

Level 1 is the best bid and offer with their sizes, plus every trade. Level 2 adds the size waiting at several prices a side, ten in Atlantive. Level 3, or market by order, shows every single order with its ID and its place in line, which is what an exact queue fill needs.

How are futures rollovers handled in a backtest?

In Atlantive, orders always trade a real dated contract. The roll happens on the day the market moved to the next contract, or a set number of days before expiry, and it is booked as a trade with its costs. Back-adjusted series serve signals and never fills.

Why would live results differ from a backtest?

Most of the gap comes from five assumptions: touch fills, guessed bar paths, stops filled at their price, missing costs and stitched contracts. Atlantive replaces each with the exchange's record. What remains is what no record holds: hidden size, how others would react to your order, your own latency and your broker's rules.

Can ChatGPT backtest a trading strategy?

A chat model can write a strategy, but it cannot fill orders against the exchange's record, so its results are a guess. Atlantive's AI backtesting agent writes the strategy in Python and this engine tests it on CME data, fill by fill.

Is Atlantive a replay tool like FX Replay?

No. Replay tools let you trade a past chart by hand, bar by bar. Atlantive tests a written strategy automatically over years of CME data, with fills from the order book, and runs a whole parameter grid in one go.

Can I trade an Atlantive strategy live?

Export an Atlantive strategy to NinjaTrader NinjaScript to run it automated, or to TradingView Pine Script to chart it and trade it through alerts. See TradingView export.

Can I see backtests of known futures strategies?

Yes. Atlantive publishes backtests of named futures strategies, from the opening range breakout to the turn of the month, each run with its published rules on CME data, one contract, after fees and slippage: its result year by year, its code, its Pine Script and a press that reruns it in your own workspace.

What is the best backtesting software for futures?

Judge it on five questions. Do limit orders wait their turn? Does it read inside the bar? Are fees and slippage charged per contract? Does the data cover your contracts and years? Does the same run give the same result? Atlantive answers yes to all five, and publishes its check against 10,146,414 real orders.

See the fills
you would get.

Bring a strategy in Python, or describe one in plain English. Atlantive runs it on the exchange's own record and shows you every fill, every tick of slippage and every dollar of cost.

No card needed.

May we measure this visit?

With your consent, Google Analytics and PostHog count visits to these pages. You can separately allow Google Ads to measure visits from our ads. Visit measurement starts only after you choose. Cookie policy