Research. Optimize. Execute. A next-generation quant operating system for systematic trading.
Joined July 2026
- Tweets18
- Following22
- Followers245
- Likes25
TheSystem Next is taking shape.
Building a serious research engine for systematic trading.
#thesystem #next
One rule: price above its 30-bar average → buy. Below → sell.
In a trending market: account +94%.
Then prices go sideways. Same rule, every signal late, every flip pays a fee: back to −7%.
The rule didn't break. The market changed.
Why not switch rules when conditions change? Because regimes are only clear looking back — a trend filter caught each change 0 to 31 bars late here.
Pairing it with a revert-to-average rule cut the worst drop from 59% to 22%. It didn't create an edge (both still ended slightly down after costs). It made the ride survivable.
Simulated prices, not market data. Educational, not financial advice.
thesystem-next.com
Backtest teardown #1
BTC 1D, long-only, 2019 → 2026:
$10k → $469k (+4,594%)
Max drawdown 35%
122 trades · 57% winners · profit factor 2.2
Same-exposure buy & hold: +243%
Looks great. Here's why we don't trust it yet.
What passed: still profitable without its 5 best trades, not carried by one lucky trade, enough trades to start.
What it is not: this is one in-sample run on the data it was built on. The number of ideas tried is not corrected for, and it is not independently verified.
Next: unseen periods, then paper trading. We'll post what survives.
Backtest, not live results. Not financial advice.
A coin that lands heads 55% of the time. Four groups of 100 traders. Same coin, same luck — only the bet size differs. After 300 flips: 2% per bet → all 100 still standing 10% → grows fastest, 7 wiped out 25% → 71 wiped out 50% → every single one Return is what you hope for. Risk is what you control.
Why a winning coin can wipe you out: lose 50%, gain 50%, and you're still down 25%. Past the best bet size (~10% here), the same edge shrinks your money — and if your edge is a little smaller than you think, the "best" bet is already too big.
Simulated coin flips, not market data. "Wiped out" = lost 90%. Educational, not financial advice.
thesystem-next.com
Strategy A wins 9 trades out of 10. Strategy B wins 3.
Per trade, A loses money. B makes it.
A: 0.9 × 1 − 0.1 × 12 = −0.3
B: 0.3 × 4 − 0.7 × 1 = +0.5
Being right is a feeling. Expected value is the edge.
Then add a toll. A fixed 0.3 per trade in fees and slippage takes 30% of A's average win — and 7.5% of B's.
The smaller your average win, the more of it the costs eat.
Simulated trades, not market data. Educational, not financial advice.
thesystem-next.com
We ran 4,096 backtests of a simple moving-average rule on pure noise — simulated coin flips with nothing to find.
The optimizer still found a peak. It looked like a discovery.
On the next window, the peak was gone.
A real edge doesn't look like a needle. It looks like a plateau — and the plateau is still there next time.
Three habits that keep a backtest honest:
1. Freeze the rules before you look.
2. Reveal the next window only once.
3. Trust plateaus, not peaks.
Simulated prices, not market data. Educational, not financial advice.
thesystem-next.com
Flip a coin that lands heads 51% of the time.
After 10 flips, you can't tell. After 100, still nothing. Somewhere around 10,000, the tilt finally steps out of the noise.
That small, repeatable tilt is what quants call edge — and it's why one great-looking backtest proves very little.
Why 10,000: a 1% tilt adds ~0.02 per flip, while the noise grows with the square root of the flips. They only separate when 0.02·n > 2·√n.
Simulated coin flips, not market data. Educational, not financial advice. thesystem-next.com
While you sleep, the engine works.
It generates strategy candidates and tests every single one — out-of-sample, under stress, across changing markets. Thousands enter every week. The gates exist to say no.
The few that pass earn a verified passport.
Research that never sleeps. You keep the proof.
This is TheSystem Next.
A research and execution platform for systematic traders.
Build and backtest ideas with AI.
Let the engine search thousands of strategy candidates 24/7.
Reject what doesn’t survive statistical testing.
Deploy only what carries evidence.
Crypto. TradFi. Research. Execution. One system.
Every strategy carries its own evidence.