SignalForgeJoin the waitlist

Beta is closed, waitlist open

Stop guessing.
Start knowing.

One workspace to watch the markets, prove an idea, hunt 30,000 more, and sell what works.

SignalForge workspaceLive

① Watch

Gold Breakout fired

XAUUSD · 4h · 2 min ago

Close broke 2,412 while the 1h trend held up. Email and push sent.

7 trackers watching · 24/7 · no code

② Prove

+$1,240

142 trades · costs in · verified

worst dip −8.2%, recovered

③ Hunt

31 markets beat this rule

XAUUSD +18.2%
NAS100 +15.4%
GBPJPY +12.9%
…28 more

scanned 30,412 in 40 seconds

30,412 markets watched

EURUSD +0.31%XAUUSD +1.04%BTCUSD −2.18%NAS100 +0.62%GBPJPY −0.44%USOIL +0.87%ETHUSD −1.33%SPX500 +0.21%AUDUSD +0.15%USDJPY −0.28%

Four tools, one loop

Everything a quant desk does.
None of the code.

01 · Trackers

It watches so
you don't have to.

Pick from dropdowns, drop in a block, drag to nest. Six conditions across three markets took four clicks, and it reads back to you in plain English.

  • nested AND / OR
  • cross-market
  • indicator vs indicator
  • own equations
  • session windows

In Python this is roughly 200 lines. Here it is under a minute, and you never see a bracket.

Gold Breakout, confirmed

ALLof the following must be true

  1. 01XAUUSD 4h close > highest high of last 30 bars
  2. 02XAUUSD 1d EMA(20) > XAUUSD 1d EMA(50)
  3. 03(ATR(14) ÷ close) × 100 < 1.8, your own equation
  4. 04DXY 1h RSI(14) < 45, cross-market
  5. 05time is inside London or New York, DST aware
  6. NOTa high-impact release within 30 min

“Alert me when gold breaks its 30-bar high on the 4-hour, while the daily trend is up and volatility is calm, as long as the dollar is soft, during London or New York, and not around a big release.”

Your tracker, read back to you. Checked for look-ahead bias before it goes live.

02 · Backtesting

Then it proves it,
candle by candle.

Spreads, commissions, drawdowns, every trade logged with 25 fields. Runs on our servers, so nobody can edit the result, including you.

Replaying history

walking through 14 Mar 2026

candles
3,128
trades
104
balance
$10,884

Yes, it made $1,240.

Fewer than half its trades won. The winners ran nearly twice the losers.

03 · OptiFind

One idea,
30,000 markets.

Lock one side of any comparison to a benchmark, then sweep everything we carry.

lock BTCUSD, then show everything outperforming it

31 results

30,412 scanned · 40s

Market6-monthvs BTC, pts
XAUUSD+18.2%+20.4
NAS100+15.4%+17.6
GBPJPY+12.9%+15.1
BTCUSD, locked−2.2%

04 · Marketplace

Sell strategies.
Sell indicators.
Keep 70%.

Publish a strategy, a custom indicator or a whole bundle. Every listing ships with a locked proof card, the engine's own record of how it performed, and your logic stays sealed inside.

Nobody can edit a performance claim here, not even the seller.

London Session Scalper
strategy · @mkadri · rated 4.6 by 128 · 34 sales
proof attached
$29
Mean Reversion Suite
3-tracker bundle · @helios · rated 4.8 by 212
proof attached
$49
Locked proof card
Sealed by the engine
Return
+12.4%
Trades
142
Max dip
−8.2%
Run ID
#000142

Buyers see the record and the caveats. They never see your conditions, and neither do we once it is sealed.

Why you can trust any of it

Nine trades never
looks like nine hundred.

Same screen, two runs. One of them is luck, and it looks like it.

7 trades · 2 weeks

90.0%

win rate

Deliberately dimmed. A great number on a tiny sample is a coin flip in a suit.

142 trades · 6 months

48.6%

win rate

Full weight. This one earned it, and it is the smaller-looking number.

"No losing trades"

Written in words, never as a fake profit factor of 100.

Sharpe

Not enough data yet. We will not print a zero and let you read it as bad.

Counted as a loss

One candle hit both exits. Unknowable, so we take the pessimistic call, and say so.

Where it works

One win rate hides
seventy-five stories.

The same rule can be brilliant in a trend and ruinous in chop. Here is where yours lives.

falling fast
falling
flat
rising
rising fast
wild
22%
38%
42%
65%
92%
normal
18%
35%
55%
72%
88%
quiet
15%
28%
45%
68%
85%

Your rule needs a trend. Skip the flat and falling columns and it keeps 84% of the profit on half the trades.

Apply and retest
75market states
8time dimensions
30K+markets to scan
25fields per trade
Blu, the SignalForge assistant

Blu can build it,
run it, or explain it.

One assistant across all four tools. Answers come with the exact numbers behind them, never advice, always evidence.

Build me an alert for gold breaking out in London
Done. XAUUSD 4h close above the prior high, restricted to 08:00 to 17:00 UTC. Want me to backtest it over two years before it goes live?
Why is my profit factor only 1.68?
You made $1.68 for every $1 lost. It is not losing often that holds you back, it is five losses over $80, all inside the March dip.
  • wins $3,182
  • losses $1,942
  • 5 trades < −$80

Where we stop

Your money never
touches our hands.

SignalForge is research software. We hold no client funds, take no deposits, and place no orders. Everything here is analysis you act on yourself, through your own broker.

No custody, ever

No wallets, no deposits, no balances. There is nothing here for us to hold or lose.

Not a broker

We do not route orders or connect to your account. You keep your broker and your control.

Next upPaper trading

Run a proven rule forward on live prices with imaginary money. Coming soon, still no real funds involved.


LaterAPI and execution

An API so you can drive your own bot from your side, and one day execution from ours, only once we are authorised, and only if you switch it on.

Priced when we open.

The beta is closed while we harden the engine. Waitlist members get in first, at the founding rate.

Free

£0

at launch

  • Alerts on a few trackers
  • 5 backtests a day
  • 1-hour candles and slower
  • Blu, limited
Join the waitlist

Power

Creators

opens after beta

  • Everything in Pro
  • Sell on the marketplace
  • Keep 70% of revenue
  • Priority backtests
  • API access
Join the waitlist

Get in first.

The beta is closed while we harden the engine. Leave your email and you are at the front of the queue when it reopens.