Real money, real orders, and it closes at 13:00 UTC today. I spent a month building a cross-sectional reversal strategy on Binance USDT-M perps, validated it in my own backtester, and switched it live on Aug 30. This is what it is holding at 10:00 UTC today. THE RULE (nothing hidden) Every day at 00–03 UTC I rank ~400 eligible USDT perps by their last 4 hours of return. I buy the 8 weakest, short the 8 strongest, equal notional. Each tranche closes exactly 12 hours later, unconditionally. Flat from 15:00 to 00:00 UTC every day. No stops on longs, a +30% disaster stop on shorts. It makes no bet on whether crypto goes up. The edge is a timing anchor, not a chart pattern: 00:00 UTC is both a funding settlement and the daily candle boundary, so mechanical flow piles up around midnight and reverses once it stops. THE OPEN BOOK — opened 01:01 UTC, closes 13:00 UTC LONG PROM +1.63% MORPHO +2.01% 4 +0.55% ZKC -0.87% ZKP -1.05% HEMI -2.71% LAB -3.12% CYS -16.71% <- the worst leg in the book SHORT SKR +12.55% <- the best leg in the book BTR +7.39% ARB +5.09% LOBSTER +2.34% NEAR +0.88% AKE -9.98% 14 legs, $5 per name, ~$72 gross. Net: -$0.09, or -0.1% of equity, 9 hours in. Yes. Flat. A 16% loser and a 12% winner in the same book cancelling each other out is not a bad day - that is the strategy working exactly as specified. It is paid by the spread between coins, not by direction. THE ACCOUNT IS $85 That is deliberate, and it is the part most people get wrong. Month one exists to measure three things no backtest can give you: real slippage per side, what fraction of orders actually fill passively, and realised net return per position. I wrote the kill lines down before I sent the first order - if slippage exceeds 10 bps per side, or net return per position drops under +8 bps, it gets switched off regardless of P&L. WHAT THE BACKTEST SAID 2021-03 to 2026-08, all costs and funding charged, marked at bar opens so there is no same-bar look-ahead: +3.05% / month compounded Sharpe 1.63 (1.51 from 2024 onward) Max drawdown -31.4% Worst month -12.4% 77% of months positive And the three controls that matter more than any of those: Random names, identical timing and size -> Sharpe -3.8 Same rule entered at 12–15 UTC instead -> Sharpe -2.3 Inverted: long the strongest -> Sharpe -3.1 If the first control had come back positive, the whole thing would have been the entry window and nothing else. It didn't. That is the test I trust. WHAT I EXPECT TO HAPPEN A -35% drawdown, and stretches of six months to a year underwater. At this Sharpe you need roughly three years of live returns to prove the edge exists. So I am not going to react to the P&L. I am going to record it, and judge the fill quality instead. I will post the closed result of this exact book after 13:00 UTC, win or lose. I built the backtesting tooling for this myself and it is free to use at virtuum-lab.com - write a rule, test it against real exchange data, see what it does before you risk anything. Not financial advice. This is a live experiment on an account small enough that losing all of it teaches me something and costs me nothing.
Bacaan kuring tina backtester kuring sorangan nyebutkeun strategi kuring kalah ku ngan saukur teu ngalakukeun nanaon.
Golden Cross, 50/200 SMA, BTCUSDT perp dina 1x, ti 2019-09-08 nepi ka kiwari. Lawan buy and hold, dina jandela waktu nu sarua.
Buy and hold: +651%, max drawdown 77% Golden Cross: +202%, max drawdown 70%
Genep transaksi salila tujuh taun, ahirna jadi leungit $44,954 ti batur anu meuli sakali sarta tuluy kabur ka luar.
Kuring unggah ieu sabab ieu perbandingan nu ampir teu aya nu ngajalankeun, sarta ngan ieu nu nangtukeun naha hiji strategi bener-bener aya hartina. Strategi mah lain keur bersaing jeung nol. Strategi bersaing jeung hal nu anjeun bakal lakukeun heula.
Di mana eta manggihan gunana, lamun memang aya gunana: drawdown. Nahan hartina ningali 77% ludes jeung teu ngajual. Rentang paling parah tina strategi nyaéta 70% — leuwih alus, tapi teu cukup pikeun ngajustifikasi genep kaputusan jeung genep taun perhatian.
Ieu hasil nyata tina alat kuring sorangan, sarta eta nyebutkeun yén strategina teu cukup alus. Backtester anu ngan bisa ngagombalkeun anjeun nyaéta cocooan.
Bandingkeun milik anjeun jeung buy and hold saméméh anjeun dagangkeun 👉 https://virtuum-lab.com
Ọ̀nà kan náà. Ìpẹ̀yà kan náà. Ìṣòro ìdánwò mẹ́wàá [five trades], ní ọjọ́ kan náà. Ẹgbẹ̀ méjì, tó yàtọ̀ sí 14 points.
Golden Cross, 50/200 SMA, 3x long BTCUSDT perp, 2020-03-25 títí dé òní, ibi kọọkan ni san èrè ìdúró (settled funding) tirẹ̀.
Binance: −16% — $9,888 ti a san fún funding Bybit: −29% — $9,641 ti a san fún funding
Àwọn ìwọlé àti ìjáde jọra gan-an. Àwọn ọjọ́ ìfọwọ́yíyàn jọra, gbogbo 5 nínú wọn. Kódà àwọn owó funding náà wà ní àyè díẹ̀ ọgọ̀rọ̀ (few hundred dollars) láàárín wọn, nítorí náà kì í ṣe ìyẹn ni ìdí náà.
Ààfiyèsí yàtọ̀ ni àwọn candle. Ẹgbẹ̀ kọ̀ọ̀kan ń tẹ́ ìye owo tirẹ̀ jáde, àti ìlànà tó ka “close above the 200 SMA” ń tú close tó yàtọ̀ díẹ̀ sí i lórí ọkọọkan. Ìlànà kan náà, ẹ̀rọ ìtẹ̀jade tó yàtọ̀, ìkún-owo (fills) tó yàtọ̀ — àti ìyàtọ̀ 14 points títí dé òpin.
Èyí túmọ̀ sí pé ìdánwò ẹ̀yẹ (backtest) jẹ́ gẹ́gẹ́ bí ojú-irin (venue) tí a ṣe e. Bí o bá dánwò lórí data ẹ̀rọ kan tí o sì rà tà lórí ẹ̀rọ míì, o ti dánwò ọgbọ́n tí o kò fi ránṣẹ́ (deploy).
Ó tún fi ààlà sí ìpamọ́ tó peye. Bí orísun data olóòótọ́ méjì bá yàtọ̀ sí 14 points lórí àwọn ìlànà tó dára gan-an, kò sí abajade backtest tí ó péye títí dé decimal point, àti ẹnikẹ́ni tó bá n sọ fún ọ pé ó péye sí ìyẹn ń ta nkan fún ọ.
Yan exchange tí o ń trade gangan 👉 https://virtuum-lab.com
"Basta gumawa ng kabaligtaran ng isang nawawalang strategy." Sige, sinubukan ko.
Golden Cross, 50/200 SMA, 3x, BTCUSDT perp, 2019-09-08 hanggang ngayon, totoong Binance funding na sinisingil.
Long lang: +359% Short lang: −100% — na-liquidate Long at short: −100% — na-liquidate
Ang inverso ng isang strategy ay hindi ito eksaktong mirror image nito, at iyon ang bahagi na hindi napapansin ng mga tao.
Hindi nagbabago ang mga gastos ayon sa iyo. Ang funding, fees at slippage ay sinisingil sa parehong direksyon, kaya ang isang strategy na kumakain ng kaunti habang long ay talo rin na *kasama* ang parehong gastos ulit kapag short. Namamana mo ang mga lugi at binabayaran mo ang toll nang dalawang beses.
At ang asymmetry ay grabe sa perp. Ang long ay puwedeng mawala ang 100% ng margin nito. Ang short ay nasa harap ng isang asset na puwedeng tumaas nang walang limit, kaya ang short-only run ay hindi lang nag-underperform — nai-liquidate ito at humihinto nang tuluyan.
Ang pagkuha sa kabilang side ng isang masamang strategy ay nagbibigay sa iyo ng ibang masamang strategy. Ang talagang kumikita ay hindi pagkakaroon ng mga costs, at tanging paraan para malaman ay ang pagsingil nang tama sa mga iyon.
Patakbuhin nang long, short, o pareho, at tingnan 👉 https://virtuum-lab.com
3x nearly doubled my return. 5x deleted the account.
Golden Cross, 50/200 SMA, long BTCUSDT perp, 2019-09-08 to today, real Binance funding charged. Same signals, same 6 trades. The only thing that changes is the size of the bet.
Look at where it stops paying. Going from 1x to 2x adds serious return. Going from 2x to 3x adds almost nothing, because funding scales with the position and eats what the leverage adds.
And 5x never gets to find out. It's liquidated 7 Jan 2022, 2 trades in, and there is no coming back from that — the remaining four signals happen to an account with nothing in it.
That's the part a return curve hides. Leverage doesn't scale your outcome, it scales your outcome until the first time it doesn't, and then the number is −100% forever.
Max drawdown at 3x is 90%. If that number doesn't frighten you, you haven't sat through one.
Find your own cliff before the market shows you where it is 👉 https://virtuum-lab.com
Same strategy. Three coins. Two accounts went to zero.
My last post showed what real funding does to a Golden Cross on BTC. The obvious next question: does the strategy work, or does BTC just go up?
So I ran the identical rules on ETH and SOL. Same 50/200 SMA cross, same 3x long, same window, real Binance funding charged. All three start on the same day, so nobody gets a head start.
BTC: −16% — survived, but with an 87% drawdown ETH: −99.9% — liquidated 3 Feb 2025 SOL: −100% — liquidated twice, the first time on its opening trade
Five trades each. Identical rules. One account limps out, two are gone.
Now drop the leverage to 1x and nothing gets liquidated: BTC +21%, ETH −38%, SOL +2%. Over six years. That's the actual edge — roughly nothing. 3x didn't amplify an edge. It just moved the liquidation price close enough to touch.
If a strategy only survives on the coin you built it on, you didn't find an edge. You fitted a chart.
Two limitations before anyone else says them: five trades is a tiny sample, and SOL's first entry doesn't come until 2023, because a 200-day SMA needs 200 days and SOL listed late.
And yes — BTC made +359% in my last post and −16% here. Same rules, same coin, same engine. The only difference is the start date: 2019 instead of 2020.
Test your rules on the coins you didn't design them for 👉 https://virtuum-lab.com
I ran the Golden Cross (50/200 SMA) on BTCUSDT perps. 3x long. Binance data, Sept 2019 → today. Ignoring funding: +2,003% Charging real Binance funding: +359% Same strategy. Same entries. Same exits. Same six trades. The gap is $96,561 in funding payments. Why it's so big: BTCUSDT funding has averaged 0.0106% per 8h settlement since 2019 — about 11.6% a year. Sounds like nothing. At 3x leverage on full equity that's roughly 35% of your account per year, and the Golden Cross holds for months at a time. You pay rent on the position the entire time you're right. It hits the downside too. Max drawdown goes from −81.6% to −90.4%. Funding doesn't just shave the top off your returns, it digs the hole deeper. Most backtest tools treat a perp like spot. No funding, no liquidation price, no margin mechanics. Then people go live at 3x and can't work out where the money went. Two caveats before anyone accuses me of selling a dream: six trades is a tiny sample, and a −90% drawdownld trade. That's the point. This is what the honest version looks like. I built a no-code backtester that charges the real funding history from the exchange you pick. Free to try check comments 👇 #BTC #Bianace
Over the last week I’ve been running a fully automated futures trading system built around a momentum/trend-following framework on the 15-minute timeframe.
The system uses multiple confirmations before entering trades, including: