Strategy research platform

Code is changing the way the stock market is traded.

Ascend Code is a strategy development machine. Write a strategy once, test it on 25 years of 1-minute data in milliseconds, prove it on data it has never seen, then run it in live simulation next to nine others.

ORB-044 Live sim
1 minute

Signal

Completed bar

Fill

Next open

Engine

Same code

25 yrs

of 1-minute bars per symbol

50M+

bars per second, per core

10

strategy families built in

10

live simulation slots

Meet The Machine

What we do, and how a strategy earns its place.

An idea becomes code, gets tested on 25 years of 1-minute data, survives a million variations, proves itself in live simulation, and only then reaches the risk gate.

The machine

Not another backtester. A loop that never stops improving.

Every strategy moves through the same cycle, and every change creates a new version instead of overwriting the old one. Nothing is lost, everything is comparable.

01

Generate

02

Test

03

Validate

04

Simulate

05

Observe

06

Modify

07

Retest

Mode 1Running

Historical backtest

FirstRate Data

Years of 1-minute history replayed through the strategy in milliseconds, with fees, slippage, stops and targets.

Mode 2Replay today, DAS next

Live simulation

DAS real-time

The same strategy code receives live bars and trades a simulated account, side by side with nine others.

Mode 3Later, behind gates

Live trading

DAS order routing

Only versions that clear every gate are promoted, behind a central risk engine with hard limits.

Build the strategy once. Change the data and the execution environment, not the strategy.

No look-ahead

Built so a backtest cannot cheat.

Most backtests that look great are quietly using information from the future. Here that is ruled out by how the engine is built, then proven by a test anyone can run.

1

Completed bars only

A strategy is handed one finished bar at a time. It has no way to reach a bar that has not happened yet.

2

Fills come later

A signal on one bar fills at the next bar's open, never on the bar that produced it.

3

Worst case inside a bar

If a stop and a target could both have been hit in the same minute, the stop is assumed to come first.

4

The truncation audit

Cut the data, rerun everything where the future does not exist, and require every feature and trade to match exactly.

Validation

From a million ideas to ten live slots.

Highest return is never the selection rule. A candidate has to survive each stage, and most do not.

1,000,000 parameter combinations
1-minute research across the full history
~1,000 candidates that survive
Walk-forward and untouched out-of-sample years
Stress worse fills, other stocks
10 strategies in live simulation

Trading research control center

Simulation
SlotVersionStrategyMode
SIM-01ORB-044Opening Range BreakoutLive sim
SIM-02VWAP-061VWAP MomentumLive sim
SIM-03MOMO-031Momentum ContinuationLive sim
SIM-04GAP-007Gap ContinuationLive sim
SIM-05MR-021VWAP Mean ReversionLive sim
SIM-06 to 10Next candidatesOpen

Illustration of the control center layout. Sign in to see live numbers.

How we work

A systematic trading business, run like software.

The edge is the process: generate more ideas than anyone can by hand, throw out almost all of them honestly, and only let capital follow what survives.

Research at machine speed

Ideas are cheap, so the system tests them by the thousand. What used to take a team a quarter runs before lunch.

Evidence before capital

Nothing is promoted on a good backtest alone. A strategy has to hold up on years it never saw and in live simulation first.

Risk is decided centrally

Every strategy trades through one risk engine: exposure caps, daily loss limits and an emergency pause that overrides everything.

Access is by invitation.

The platform is private. Leave your email and we will be in touch, or sign in if you already have an account.

Sign in

We use your email to review and reply to this request. Read our Privacy Policy and Website Terms.