After 6 months of intensive development, shitty code, sacrificed goats and dancing around fire, our team finally gave birth to an algo that bears enough valuable updates (not just weekly adjustments) to get a full version step-up.
Meet SAF3 — the upgraded rollercoaster that's leaving our previous SAF2 system in the dust. Or not. We'll find out. Technically it is not supposed to leave it in the dust because the ride is expected to be smoother, not faster - same velocity but less sharp turns.
SAF3 Supersedes SAF2 Across Every Key Metric
Hopefully not drawdown.
As many of you know, we always thought SAF2 is something special — and clearly it is a "very special" algo. It generated good profits on our personal accounts, our mutual account, and our clients' accounts. But markets evolve, and so must the algorithms.
SAF2 has been deviating slowly but surely from prior results — not that it lost money recently, but we could see deviations from expected baselines. We kept adjusting it on a weekly basis. Sometimes the algo would be adjusted multiple times per week: new features, new symbols, new timeframes, new logic, etc. Not because it deprecates fast, but due to everchanging markets which we need to adjust to. Nevertheless, none that we would deem a game-changer that would earn a next-gen upgrade.
Live Testing Starts This Week
This week (sorry, it will not be a logical start date like Monday or 1st of the month) we will be launching SAF3 on a mini-account with ~$1,000 USD to test-run it in the real world.
Important Disclaimer
NB: The following metrics are based on live data of SAF2 and BACKTEST of SAF3.
The backtest was done as granular as possible, with artificial slippage, commissions, and even funding fees — but not live, so not real YET.
The Numbers (Backtest vs Live Comparison)
RISK MANAGEMENT EVOLUTION
SIGNAL QUALITY ENHANCEMENT
OPERATIONAL IMPROVEMENTS
What Makes SAF3 a Game-Changer?
Nothing yet.
The latest updates that are the most crucial have not been evaluated in the REAL MARKET — we still keep it real and honest.
Main Difference
SAF3 has additional filtering systems implemented, hence our expectations are:
- Less trades
- Smoother sailing
- Smaller and gentler drawdowns
- Not so many trades at the start of trends
- Faster recovery
If it survives our live testing phase, we expect to see:
1. Signal Quality Over Quantity
SAF2's approach was relatively simple compared to the new one: identify range-bound conditions and trade most opportunities, just because of mathematical edge it would eventually pull us up.
With SAF3 we've eliminated approximately 41% of SAF2's trades — but not so many profitable ones got cut off, at least not in the backtest.
Translation: Fewer trades, but the ones we take should be higher quality.
2. Volatility Taming
Something we've been trying to get for a while — not that we didn't trust the system, but even with automation it still might be a bit pressing on the psyche.
SAF3's trend filters prevent it from instantly jumping against strong directional moves. Instead, it will jump in later, hopefully.
Translation: Less emotional stress, smoother equity curves.
3. Recovery Power
This is a very important part of any low win-rate strategy. The faster the recovery, the better your sleep — because you get your cash back with extras faster.
Especially as we are unable to use compounding and instead have to resort to fixed step-based scaling.
Translation: When we hit drawdowns, we should climb out faster.
Launching SAF3 for the Public
We will be giving it ~$1,000 USD to trade with.
Depending on:
- The amount of trades it takes (sample size)
- The market conditions during testing
We will decide on the duration of the testing phase before it goes live.
Transparency Commitment
We will of course share all the results publicly and have them updated even more frequently than SAF2 for transparency purposes. We will do:
- PnL calendar
- Individual trades visible somewhere
- Daily updates
We might not use position consolidation because doing it on Binance is a bitch and might cause quite noticeable discrepancies, especially on a small trading account.
But Daily PnL is taken from realized profits endpoint and will be 100% accurate, same for open/close SL/TP, position size — those are all always 100% correct.
It's just the realized profit calculation that might be off by a few percent of individual closed trades, not daily.
The Reality Check
We're comparing live trading results (SAF2) with backtest results (SAF3).
Even though we add random artificial slippage, commission, and even funding fees to the backtest:
We will believe it when we see it.
That's why we're starting with real money, real trades, and real transparency.
Stay Tuned
Pray for SAF3 and our ~$1,000 USD that we hand it over.
This isn't marketing theater. This is us putting our money where our mouth is — again.
If SAF3 performs as expected, you'll see it in the numbers.
If it doesn't, you'll see that too.
No edits. No cherry-picking. No bullshit.
P.S.
The other test strategy is running somehow. We're overloading it to the max for now to see what the worst-case scenario looks like. We'll share an update on X, Facebook, and the blog soon enough.