Edge Lab

Finding an edge doesn’t have to be rocket science.

Five strategies that already work. Pick one, press optimise, and Edge Lab tests every version of it against years of real sessions to find the version that holds up on your instrument and your session. Then you copy the settings straight into an indicator.

Start free See the three steps
Included with Pro · no course, no signal group
console.tradestar.app / market-edge / lab

Edge Lab.

Sweep a strategy’s settings against real sessions, and see which held up.

New sweepBeta
Held up

A 30 minute opening range, taken both ways.

5,184 combinations tested on 1 year of Nasdaq 100 New York sessions. 1,204 cleared the constraints, and this one sits in the middle of the widest working zone on the grid.

Net over 1 year+$41.3KNet of commission and slippage
Win rate41.8%
Profit factor1.62
Max drawdown−$6.4K
Unseen dates+1.9pp
Trade it like this30 minrange5mchartboth waysdirection0.70×stop1.75×target
Open the full reportStrategy settingsSave to playbookOr compare it against the other 19 below.

The best of what cleared.

Top 20 of 1,204 that cleared · map the zone, or compare the rows
Response surfaceRanked list
The choice

Two paths, one shows its working.

The guru model

A course, a Discord, and one set of settings that came from nowhere.

One structure, renamed and resold, for $500 to $1,000.
Settings handed down with no test behind them.
The same numbers for every instrument and every session.
Screenshots as evidence, and no sample size anywhere.
Edge Lab

Five strategies that already work, and a system that finds your version of one.

Five documented structures, each with its own swept parameters.
The system sweeps every version and reports what survived.
Settings measured on your instrument, session and timeframe.
Copy the recipe into an indicator. Included with Pro.

Every number on this page came from a test you can rerun yourself.

How it works

Three steps, start to chart.

01

Pick a strategy.

Choose one of the five, then the instrument, the session and how far back to test. Three decisions on one screen. Everything else has a sensible default, and every dial is still there if you want it.

Five to choose from.
Your instrument, your session.
Starter, Balanced or Wide.
console.tradestar.app / market-edge / lab / new sweep
StrategyORB Opening range break
Asset & tickerNAS100 · Nasdaq 100
SessionNew York
Period1 year
Run the sweep
02

Press to optimise.

The system tests every combination of stop, target, entry and filter against up to two years of real sessions, then re-runs the whole search across five rolling windows and judges each window's winner only on the dates that came after it. That part is not optional and you cannot switch it off.

Thousands of versions, scored the same way.
Five walk-forward windows, not one.
Monte Carlo on the survivor.
console.tradestar.app / market-edge / lab
Every algo · every ticker · every setting

Find the version of your strategy that actually held up.

01Set it up

Pick the instrument and the algo in the rail on the left. A preset picks the grid for you.

02Sweep it

Every combination runs against real sessions, net of commission and slippage.

03Read the answer

The one that survived, in plain English, with the full report behind it.

Sweep 11,520 combinationsRoughly 20 seconds. You can change anything and run it again.
Find me the most profitable setup637 sweeps ran at the weekend - every strategy, every market, every session.
Start hereNew here? Sweep one we already found. Initial balance break on S&P 500.64.2%Set it up
03

Copy the settings.

You get one recommendation in plain English, the numbers behind it, and a settings sheet. Copy or download it, set an indicator to those exact values, and the strategy is drawn on your chart the way it was tested.

A sentence, then the numbers.
One export, copy or download.
Saved to your playbook.
console.tradestar.app / market-edge / lab / result
Held up

A 30 minute opening range, taken both ways.

5,184 combinations tested on 1 year of Nasdaq 100 New York sessions. 1,204 cleared the constraints, and this one sits in the middle of the widest working zone on the grid.

Net over 1 year+$41.3KNet of commission and slippage
Win rate41.8%
Profit factor1.62
Max drawdown−$6.4K
Unseen dates+1.9pp
Trade it like this30 minrange5mchartboth waysdirection0.70×stop1.75×target
Open the full reportStrategy settingsSave to playbookOr compare it against the other 19 below.
Why our winner is the winner

A leaderboard hands you the luckiest setting.

Rank a sweep by profit and the top row is usually one bright cell surrounded by cold ones. It made the most money in the test and it will not do it again, because nothing either side of it worked.

Edge Lab looks for the plateau instead, a block of settings where the neighbours work too. A version that only pays at exactly 0.6× stop is fitted. One that pays from 0.5× to 0.8× is real, and it is the one you can trade without standing on the edge of a cliff.

Net P&L by stop × targetHow to read a surface
0.4×0.5×0.6×0.7×0.8×1.0×
✕
○
1.0×1.25×1.5×1.75×2.0×2.5×3.0×
○What Edge Lab recommends. The middle of a block where every neighbour is profitable too.
✕What a leaderboard would print. The best single number in the test, cold on all four sides.
Walk-forward

The search is re-run five times, and judged on what came next.

Slicing one already-chosen setting's trades by date is not out-of-sample: those dates picked the setting. So each of the five windows re-runs the whole optimisation over its own training dates, then measures the winner that fit produced on the dates immediately after it. No test date ever contributed to the fit being judged on it.

Permutation null

Priced for the size of the search, not one lucky cell.

A result is compared against the distribution of the best cell a search this wide would find in shuffled data, not against one pre-chosen cell. Run the naive version through a 200-cell grid on pure noise and it declares significance 200 times out of 200. Correcting for the search is the difference between a p-value and a decoration.

Sample size

Under thirty trades, both checks refuse to answer.

A p-value across eleven trades is a confident-looking figure with nothing behind it. Below thirty trades, or fewer than three usable windows, the honest output is that the sample cannot support the claim, not a number with a caveat beside it.

The last step

The result is a setting, not a report.

Research that ends in a PDF changes nothing. A sweep ends in a configuration. The run exports as a settings sheet, every row named for what the algo actually did rather than the field it was stored in, so an indicator set to those values draws the strategy exactly as it was tested.

See the indicators
Opening range break · Nasdaq 100 · NYCopy settings
Range30 min
Chart5m
Entry window60 min
DirectionBoth ways
Stop0.70×
Target1.75×
Only recommended once the walk-forward windows agree.

Trade the version that held up.

Pick a strategy, run the sweep, copy the settings. The first one takes about five minutes.

Start free See pricing
[ trade★star ]Evidence for every serious trader. Broker sync, prop tracking, and edge stats in one workspace.
© 2026 Tradestar. Trading involves risk of loss.Historical stats, not predictions. Trade your own plan.