Skip to main content
Koval
Koval — Ukrainian for blacksmith

Backtest trading strategies without writing code

Build strategies visually, then replay years of market data bar by bar — every fill and fee included. Test with paper funds in an exchange sandbox, use the results to refine your strategy, and keep trading decisions yours.

Join the waitlistRead the engine source

Waitlist open · free plan at launch · paper & sandbox only

Get early access

Execution boundaries — implemented today

Binance

futures testnet

WhiteBIT

fail-closed adapter

+ pluggable adapters

open adapter interface

Why Koval

Forge your strategy. Test it against reality.

Koval does not pretend a backtest can predict the market. It replays the rules you forged bar by bar over historical OHLCV data, with every evaluated bar, simulated fill, and fee open for inspection — evidence you can challenge before moving to paper or sandbox testing.

ledger — what you get5 / 5

[✓]

01

Forge your strategy visually — craft rules as a node graph, not a script

[✓]

02

Deterministic backtests — same inputs, same result, every run

[✓]

03

Fees-only cost model, stated up front — no hidden slippage assumptions

[✓]

04

Every bar, fill, and fee inspectable after the run

[✓]

05

Open-source engine — MIT core on GitHub, readable before you trust it

Workflow

From blocks to a sandbox-tested strategy in four passes

01

Build

Drop market data, indicators, conditions, and actions on the canvas and wire them together. The graph is the strategy — there is no hidden layer underneath it.

02

Backtest

Replay years of OHLCV data bar by bar with an explicit fees-only cost model. Read return, max drawdown, win rate, and profit factor — then read the trades that produced them.

03

Paper / sandbox

Push the same graph to Binance futures testnet with paper assets and watch it process live candles in real time. WhiteBIT stays fail-closed until a verified non-money sandbox exists.

04

Capture & compare

Line the next run up against the last — Sharpe, Sortino, and the trades behind them. Backtests are deterministic, so what changed is what you changed. Rewire a node and replay: no code, MIT-licensed engine underneath.

Features

Everything on the bench, nothing behind glass

Visual strategy builder

Drag indicators onto a canvas and wire the rules. The node graph is the single source of truth — readable by you, executable by the engine.

Deterministic backtesting

Bar-by-bar replay over real historical OHLCV sourced from exchanges, with an explicit fees-only cost model — no generated data or AI hallucinations. Use the same engine version, dataset, and settings, and every rerun produces exactly the same result.

Graph-native indicators

EMA, RSI, ATR and friends are nodes you compose and rewire, not fields in a settings form.

Sandbox execution

Run the same graph against supported exchange sandbox APIs with paper funds. Observe how each revision behaves on live candles, uncover weak logic, and refine the strategy in a sandbox-only environment.

Pluggable adapters

Binance futures testnet execution and the WhiteBIT contract boundary are implemented. WhiteBIT stays fail-closed because it has no public sandbox or testnet.

Full trade ledger

Every order, fill, and fee is written down. Audit any run, or export it and check the math yourself.

Transparency

Read the run, not just the result

Sample data — illustrative

strategy graph
EMA 50EMA 200cross ↑BUY
backtest metrics

Return

+18.4%

Drawdown

-9.6%

Win rate

54.2%

Trades

192

equity vs benchmark
strategy +18.4%buy & hold +9.1%
trade history
14:02BUY0.042 BTC @ 67,412.50
17:35SELL0.042 BTC @ 67,911.00+20.94
09:14BUY0.038 BTC @ 66,988.25
11:47SELL0.038 BTC @ 66,540.10-17.03
15:22BUY0.041 BTC @ 66,712.00
Paper run — EMA Cross 50/200
+2.1

+$312.09

BTC/USDT 1h · Binance sandbox · sample

paper
running

uptime 3d 06:12

sandbox order log
12:04:11 connect binance-sandbox … ok
12:04:12 subscribe BTC/USDT 1h
13:00:02 signal cross↑ → BUY 0.042
13:00:02 order accepted #98F2-11
13:00:03 fill 0.042 @ 67,412.50 fee 0.10%
14:00:01 hold — no signal

Backtesting

Prove it before you run it

Sample backtest — illustrative, fees-only model

Total return
+9.3

+18.4%

vs. buy & hold · sample

Max drawdown

-9.6%

worst peak-to-trough

Win rate

54.2%

104 of 192 trades

Profit factor

1.42

gross win / gross loss

Equity curve

2023-01 → 2024-12 · BTC/USDT 1h

+20%+10%0%-10%2023-012024-012024-12strategybuy & hold

Trade PnL distribution

-4%0+4%

Monthly returns

2023

2024

Built in the open

The engine is open. The numbers are checkable.

Open-source engine

The engine that runs your strategies is public and MIT-licensed. Read exactly how a backtest is computed before you trust a number it prints.

github.com/koval-finance/koval-engine

Sandbox-first

There is no real-money mode to turn on. Koval talks to paper accounts and exchange sandbox APIs, and nothing else.

Paper & sandbox only

Nothing hidden

Backtests are reproducible and every bar, fill, and fee is inspectable after the run.

fees-only cost model

What is public — read it yourself

koval-engineMIT

The engine that runs every backtest.

Read the bar-by-bar backtest loop, the fees-only cost model, and the exchange adapters that produce every number the app shows you.

The Backtrader adapter, kept separate.

Isolated in its own repository so its GPL-3.0 terms stay off the MIT core. Read the adapter boundary for yourself.

The hosted application built on top of these repositories — the visual builder, accounts, and managed runs — is proprietary and is what a subscription pays for.

Pricing

Free to start. Priced at launch.

Pro, monthly

Price announced at launch

The full visual strategy builder: create your own strategies, run unlimited backtests, paper trade, and test sandbox execution.

Build your own strategies, no code

Unlimited backtests on historical OHLCV

Paper trading with simulated capital

Binance futures testnet execution

Compare runs side by side

Billed monthly. Cancel any time.

Free

Default

$0

Available when the app opens

Register, run one preset strategy end to end, and read the full backtest: every bar, every fill, every fee. Enough to judge Koval before you pay for it.

Join the waitlist

One preset strategy

Full backtest report

Fees-only cost model, stated up front

No card required

Pro, annual

Best value

Price announced at launch

Everything in Pro, billed once a year at a lower effective rate.

Everything in Pro monthly

Lower effective monthly rate

Waitlist members get the launch price first

Prices are not set yet. We are still sizing a fair monthly cost against real infrastructure costs. Waitlist members hear the numbers first.

Early access · Koval waitlist

Get ready to forge your next strategy

One email when the forge opens. Nothing else.

Required
Optional
Optional

No spam. No sharing. One launch email.

FAQ

Straight answers

It is software for designing rule-based trading strategies through a visual interface instead of writing code. Koval uses connected blocks so traders can define signals, filters, risk rules, and execution behavior clearly.

Ready at the anvil

Join the waitlist

Koval

open-core · MIT · v0.9.0

Koval executes strategies against paper accounts and exchange sandbox environments only. It never places real-money orders. All performance figures shown on this page are samples for illustration and are not indicative of future results.