Learn · Updated September 2026

Systematic crypto strategies

People searching this phrase want a written rule they can replay — not a top-10 RSI list. This page is that: 1D templates, a size calculator, a DCA ladder, and a backtest on real Binance.com daily bars. Strategy A is Pi Slice. B is Band Fade. C is OTT Fade.

01Read the specBook, clock, enter, exit
02Size the ticketCapital, slices, ladder
03Replay the book1D, 4h, or 1h bars

What “systematic” means here

Systematic is the opposite of discretionary. The computer (or you, by hand) does not get a vote after the rule is written. If you skip a buy because the candle “looks scary,” you left the system.

Search results for this query are Quantpedia-style writeups, academic papers, and rule-based blogs — momentum, mean reversion, a volatility filter. That is the job of this page. The other learn article covers broader crypto strategy types. This one is the lab: 1D, measurable, replayable.

Crypto trading strategies (overview) · Paper trading

Strategy A — Pi Slice

Sixty equal tickets. Buy the most oversold blue chip on RSI of the selected TF (1D, 4h, or 1h). Take π. That is the name — not a promise.

Written rule

Pi Slice

Harvest short oversold bounces in a fixed blue-chip book. Do not marry a bag.

Book: Top 25 liquid Binance.com USDT names (fixed list, not live rank). If none print RSI ≤ 30, widen to 50. If still none, sit out.
Clock: 1D, 4h, or 1h close. One scan per bar. Calendar limits stay calendar: 2 DCAs per UTC week, 1 drip per month.
Enter: Lowest RSI at or under 30 among names you do not already hold. Fill at that bar’s close. One new name per bar. Size = the current ticket (starts at $200, then rolls).
Size: First ticket $12,000 / 60 = $200. After each take-profit close, leftover after tax and self-dividend grows the next ticket until the max first ticket (default $500). Default: reset the cycle to $200 so cash remains for DCA. Compound mode lets the ticket keep growing. You cannot spend cash you do not have. Optional overlays (off by default): Vol Note scales that ticket from BTC realized vol; Drawdown Gear cuts it as the desk falls from its own peak.
Adds: Last fill, not average. Bar low 5% under the last buy. At most two such DCAs per UTC week, each using the current ticket. Prefer the oldest eligible lot — unless that lot already has many adds and one ticket would barely move it, then pick the lot where the ticket still matters. No add on a bar you opened a new name.
Bag brake: If close is 30% under the first buy, regular DCA stops. That name moves to drip: one add per calendar month, size = 1% of money already in it.
Exit: Bar high tags take-profit % above average entry (default 3.14%). Fill at the target. Then roll the ticket.
Dead when: Tickets are gone and the book is all bags. You drip. You do not invent an 8th add. You do not “just this once” buy a name you already hold.

Lineage — what this borrows, what it does not

J. Welles Wilder Jr., 1978 — New Concepts in Technical Trading Systems. RSI; 30 is his oversold line. We buy only at or under 30, and we take the lowest RSI in the book on that bar.
Narasimhan Jegadeesh, 1990 — “Evidence of Predictable Behavior of Security Returns,” Journal of Finance. Short-horizon cross-sectional reversal: recent losers often bounce more than they continue. Buying the weakest oversold name in a basket is that idea. It is an equity paper, not a crypto proof.
Equal-unit tickets — Van Tharp’s unit sizing, and the older habit of betting in equal chips. $12,000 ÷ 60 = $200. One new name = first ticket. After a close, the ticket rolls. You do not size from a feeling.
π (3.14%) is a named target — not a mathematician’s law of Bitcoin. π is how you remember the number. 3.14% must clear two fees and leave a sliver. That is the only math.

Replay it in the Pi Slice backtest below. That run is a history walk, not a live desk.

Pi Slice — basket backtest

Replay desk

Pulls the selected TF for the fixed blue-chip list (not live market-cap rank), then walks one bar at a time. Same rules as 1D: RSI 14, last-fill DCA, 2 DCAs per UTC week, 1 drip per calendar month, TP from average. New buy first. No regular DCA or drip on a new-buy bar. Same-bar fills — optimistic versus a live book. 4h / 1h load more candles — start with 1 year.

Clock

TimeframeHistory

Ticket

CapitalSlicesMax first ticketWhen ticket hits max

Costs

Fee % / fillTP %Tax %Self-dividend %

Fat first buys starve DCA. Reset (default) starts the ticket over at capital ÷ slices when it would reach the max. Compound ignores the max and lets the ticket keep growing.

Money overlays — off is the original shop

Vol Note — size the ticket from BTC realized vol vs a 50% annualized target. Double vol → half note. Still the same RSI buy.Drawdown Gear — cut the note from the desk’s own peak: 100% / 50% / 25% / $0 at −6% / −12% / −20%. Stays at $0 until drawdown is back under 6%. Drip also pauses at $0.

Widen to top 50 if nothing in the 25 is under RSI 30 (loads 50 names)

Strategy B — Band Fade

Fade a 3.5σ stretch on the 1h book. Both sides. Take 1% or sit with the bag. Capital rules stay Pi Slice. The band is the entry — not a new money system.

Written rule

Band Fade

Buy panic through the lower band. Short mania through the upper. One percent and out.

Job: Harvest a snap-back after a 3.5σ 1h stretch. Do not ride the trend. Do not invent a stop.
Book: Top 25 liquid Binance.com USDT names only (same fixed list as Pi Slice). No widen to 50. Max 5 open names at a time.
Clock: 1h bars. Bands from the prior 20 closes (20, 3.5σ). Fill on this hour’s high/low. One new name per hour — the largest stretch. Calendar limits stay calendar: 2 DCAs per UTC week, 1 drip per month.
Enter: Long when the hour trades 0.50% through the lower band (fill at lower × 0.995). Short when it trades 0.50% through the upper (fill at upper × 1.005). Skip the hour if both print. Skip a name you already hold. Skip if 5 names are already open.
Size: Same ticket as Pi Slice: $12,000 / 60 = $200 first note, then roll after a winner (tax, self-dividend, reset/compound). Vol Note and Drawdown Gear are on by default. Leverage is optional; 1× spends cash like Pi Slice. Above 1× the ticket is margin.
Adds: If the lot is underwater but not yet a 30% bag: last-fill DCA, 5% against the last print (long: lower; short: higher). Same 2-per-UTC-week cap. No add on a new-name hour.
Bag brake: Long close 30% under the first fill, or short close 30% above it: regular DCA stops. Drip begins the same way as Pi Slice — 1% of margin, once per calendar month, at the close. No stop-loss.
Exit: Take-profit 1% from average. Long: hour high ≥ avg × 1.01. Short: hour low ≤ avg × 0.99. Fill at the target. Then wait for the next stretch — you do not re-enter the same name on that hour.
Dead when: Five bags, cash gone, gear at $0. You drip. You do not open a sixth name. You do not “just this once” fade a 2σ print.

Lineage — what this borrows, what it does not

John Bollinger, 1980s — bands as a volatility envelope, not a forecast. 2σ is his default. 3.5σ is rarer. We wait for another 0.50% through that 3.5σ line so the fade is a stretch, not a tap.
Mean reversion, not breakout — buying 0.50% above the upper band would chase the move. This desk fades: long through the floor, short through the ceiling. If you wanted the breakout, write that rule. Do not flip this one mid-week.
Pi Slice capital — last-fill DCA, 30% bag drip, ticket roll, Vol Note and Drawdown Gear on by default. The band picks the name. The shop still sizes the ticket.
1% is a named target — not a law of 3.5σ. It must clear two fees and leave a sliver. Leverage is optional and it also leverages the hole. No stop does not mean “no risk.”

Replay it in the Band Fade backtest below. That run is a history walk, not a live desk.

Band Fade — 1h backtest

Replay desk

Pulls 1h bars for the fixed top-25 book, then walks one hour at a time. Bands from prior closes (no same-bar peek). Long 0.50% through the lower band, short 0.50% through the upper. TP 1% from average. No stop. Max 5 names. Same last-fill DCA, weekly cap, and monthly drip as Pi Slice. Same-bar high/low fills are optimistic versus a live book. Start with 90 days.

Clock

TimeframeHistoryBB periodBB σ

Ticket

CapitalSlicesMax first ticketWhen ticket hits max

Entry / exit

Through band %TP %Max positionsLeverage (optional)

Costs

Fee % / fillTax %Self-dividend %

Leverage 1× spends cash like Pi Slice. Above 1×, the ticket is margin; notional = margin × leverage. No stop. A 3.5σ stretch that keeps going still drips after −30% from the first fill.

Money overlays — on by default; uncheck for raw tickets

Vol Note — size the ticket from BTC realized vol vs a 50% annualized target. Double vol → half note. Same band entry.

Vol lookback (bars)Target ann. vol %

Drawdown Gear — cut the note from the desk’s own peak: 100% / 50% / 25% / $0 at −6% / −12% / −20%. Stays at $0 until drawdown is back under 6%. Drip also pauses at $0.

Strategy C — OTT Fade

Fade a 5% stretch away from Harvest OTT 1000 when the 200 / 500 / 1000 stack is live. Both sides. Take 0.75% or sit with the bag. Capital rules stay Pi Slice.

Written rule

OTT Fade

Short the up-stack blow-off. Buy the down-stack dump. Do not chase the Harvest stack.

Job: Counter-trend scalp when price is 5% or more beyond OTT 1000 on a live Harvest stack. Harvest itself buys the reclaim after a dip through 1000. This desk does the opposite job: fade the extension.
Book: Top 50 liquid Binance.com USDT names (Pi Slice top 25 plus the widen-to-50 list). Fixed list, not live rank. Max 10 open names at a time.
Clock: 3m, 5m, 15m, 30m, or 1h. OTT 200 / 500 / 1000 from prior closes (VAR, 5% — same as OTT Harvest, not Co-Pilot 200 / 500 / 750). Adjacent lines closer than 0.15% of the largest OTT are mixed, not a stack. One new name per bar — the largest stretch. Calendar limits stay calendar: 2 DCAs per UTC week, 1 drip per month.
Enter: Short when OTT 200 > OTT 500 > OTT 1000 and the bar high tags 5% or more above OTT 1000 (fill at OTT 1000 × 1.05). Long when OTT 1000 > OTT 500 > OTT 200 and the bar low tags 5% or more below OTT 1000 (fill at OTT 1000 × 0.95). Skip mixed / unknown. Skip a name you already hold. Skip if 10 names are already open.
Size: Same ticket as Pi Slice: $12,000 / 60 = $200 first note, then roll after a winner (tax, self-dividend, reset/compound). Vol Note and Drawdown Gear stay off unless you switch them on. Leverage is optional; 1× spends cash like Pi Slice.
Adds: If the lot is underwater but not yet a 30% bag: last-fill DCA, 5% against the last print (long: lower; short: higher). Same 2-per-UTC-week cap. No add on a new-name bar.
Bag brake: Long close 30% under the first fill, or short close 30% above it: regular DCA stops. Drip begins the same way as Pi Slice — 1% of margin, once per calendar month, at the close. No stop-loss.
Exit: Take-profit 0.75% from average. Long: bar high ≥ avg × 1.0075. Short: bar low ≤ avg × 0.9925. Fill at the target. Then wait for the next stretch — you do not re-enter the same name on that bar.
Dead when: Ten bags, cash gone. You drip. You do not buy the 5% stretch on an up stack. That is a chase. Write that rule separately if you want it.

Lineage — what this borrows, what it does not

OTT Harvest stack — 200 / 500 / 1000, VAR, 5%, 0.15% minimum step. Same lines the harvest scanner prints. Co-Pilot stays on 200 / 500 / 750. Do not mix the two books.
Counter-trend, not Harvest — Harvest buys a dip through OTT 1000 and the reclaim. Buying 5% above OTT 1000 on an up stack chases the extension. This desk shorts that print and buys the mirrored dump.
Pi Slice capital — last-fill DCA, 30% bag drip, ticket roll. Overlays off by default. The stack picks the name. The shop still sizes the ticket.
0.75% is a named target — two 0.1% fees leave about 0.55% before tax. Tight. A stretch that keeps going becomes a bag. No stop does not mean “no risk.”

Replay it in the OTT Fade backtest below. That run is a history walk, not a live desk.

OTT Fade — backtest

Replay desk

Pulls the selected TF for the fixed top-50 book, then walks one bar at a time. Harvest OTT 200 / 500 / 1000 from prior closes (no same-bar peek). Short a live up stack 5% above OTT 1000; buy the mirrored down-stack dump. TP 0.75% from average. No stop. Max 10 names. Same last-fill DCA, weekly cap, and monthly drip as Pi Slice. OTT 1000 needs ~1,000 closed bars — 3m and 5m use shorter windows so the download stays sane. Same-bar high/low fills are optimistic versus a live book.

Clock

TimeframeHistoryStretch vs OTT1000 %Min stack sep %

Ticket

CapitalSlicesMax first ticketWhen ticket hits max

Entry / exit

TP %Max positionsLeverage (optional)

Costs

Fee % / fillTax %Self-dividend %

Leverage 1× spends cash like Pi Slice. Above 1×, the ticket is margin; notional = margin × leverage. No stop. A 5% stretch that keeps going still drips after −30% from the first fill.

Money overlays — off is the original shop

Vol Note — size the ticket from BTC realized vol vs a 50% annualized target. Double vol → half note. Same stack entry.Drawdown Gear — cut the note from the desk’s own peak: 100% / 50% / 25% / $0 at −6% / −12% / −20%. Stays at $0 until drawdown is back under 6%. Drip also pauses at $0.

Four 1D templates you can run now

Starter rules so the tools are not empty. Long-only. Spot. One timeframe. Change the numbers in the backtest — do not treat the defaults as “the edge.”

1. Daily SMA cross

Stay long only while the faster average is above the slower one.

Enter: 1D close while fast SMA is above slow SMA (first bar both averages exist, or a cross back up).
Size: All cash, minus fee. One position. No adds.
Exit: Fast SMA is back below slow SMA. Fill at that day’s close.
Dead when: Chop — many crosses, fees eat the edge. Pause if you get 6+ round trips in a quarter with a falling equity line.

2. Donchian breakout

Buy strength after a new N-day high. Sit out the range.

Enter: 1D close above the prior N-day high (today’s bar does not count in the channel).
Size: All cash, minus fee. One position.
Exit: 1D close below the prior N/2-day low.
Dead when: A tight range after a spike. If the last 3 breakouts all stopped inside 10 days, sit out.

3. RSI fade to the mean

Buy a washed-out daily RSI and exit when it normalizes.

Enter: 1D RSI at or below the buy level (default 30).
Size: All cash, minus fee. One position. No averaging the fade.
Exit: RSI at or above the sell level, or close back to the SMA — whichever first.
Dead when: A trend day that keeps printing new lows. One fade. If it is still red at the SMA, you already exited.

4. Dip DCA (from last fill)

Accumulate a dip with a capped ladder, then take a small target from average entry.

Enter: Base at the day’s close when flat. Next buy when the daily low reaches N% below the last fill.
Size: Fixed base USD + safety count + volume scale. Cap is the full ladder — not “one more.”
Exit: Daily high reaches take-profit % above average entry. Fill at the target.
Dead when: Flash crash through the last safety with no bounce. The open cycle marks to market. Do not add a 7th safety by hand.

Calculators

Size the risk before you replay. The ladder shows max capital if every safety fills — that number is the real position, not the base order.

Pi Slice ticket (rolling)

First ticket = capital ÷ slices. After a take-profit close, fees come out, then tax on net profit (approx, not a filing), then the self-dividend. The leftover is added to the next buy.

Capital (USDT)SlicesFee % / fillTP %Tax %Self-dividend %

First ticket $200.00 · net after fees $6.07 · tax $1.52 · dividend $2.28 · growth $2.27 · next ticket $202.27

Position size

How many coins if you risk a fixed % of the account to a named stop. Long-only. No leverage.

Account ($)Risk %EntryStop % below entry

Risk $100.00 to a stop at 95. Size 20 coins — notional $2,000.00.

Dip-DCA ladder

Same math as the public calculator: deviation from the last fill, optional volume scale, take-profit from average entry.

Start priceBase order ($)Safety ordersDeviation %Volume scaleTake-profit %

Full ladder $350.00 · avg 94.04 · last fill 88.58 (11.42% down) · TP 95.45 · bounce from last fill +7.76%

Step	Price	Order	Avg
Base order	100	$50.00	100
Safety #1	98	$50.00	98.99
Safety #2	96.04	$50.00	97.99
Safety #3	94.12	$50.00	96.99
Safety #4	92.24	$50.00	96
Safety #5	90.39	$50.00	95.02
Safety #6	88.58	$50.00	94.04

Full DCA calculator

1D backtest

Default fee is 0.1% per fill (spot-taker ballpark). Dip-DCA fills safeties at the ladder price when the daily low reaches it, and takes profit at the target if the daily high tags it. Trend rules fill at the close. That is optimistic versus a live book.

1D backtest (browser)

Pulls Binance.com spot daily candles and replays the selected rule. Fee on every fill. Open trades mark to the last close. Not Studio, not live.

PairHistoryStart equity ($)Fee % / fill

Fast SMASlow SMA

Want the bot engine instead? Create an account and open Backtest Studio after paper.

FAQ

What are systematic crypto strategies?

A written rule that runs the same way every time: same entry, same size, same adds, same exit. You do not decide from a feeling on the red candle. A 1D SMA cross, a Donchian break, an RSI fade, and a capped dip-DCA ladder are systematic. “I like this dip” is not.

Why daily (1D) and not 5-minute?

Daily bars cut noise and chat. One decision a day is enough to see if a rule has an edge. The four templates stay on 1D. Pi Slice can also replay on 4h and 1h with the same calendar limits (2 DCAs per UTC week, 1 drip per month) — a faster clock, not a new rule. Band Fade is 1h only (Bollinger 20, 3.5σ). OTT Fade replays 3m, 5m, 15m, 30m, and 1h. OTT 1000 still needs about 1,000 closed bars, so faster clocks use shorter history windows.

Is this backtest the same as Tradeeon Backtest Studio?

No. This lab replays four long-only 1D templates in your browser on Binance.com daily closes (spot). Studio uses the same condition engine as a live bot, more timeframes, and needs an account. Use this page to reject a broken daily rule. Use Studio when you want the bot’s exact logic.

Can I lose money following these rules?

Yes. Every template here can and will lose. Fees are charged. An open trade at the last bar is marked to market, not magically closed. A green backtest is not a live result. Paper on live prices at the size you would actually use before any key.

What is Pi Slice?

A basket rule on 1D, 4h, or 1h: start with 60 tickets, scan a blue-chip book, buy the lowest RSI at or under 30 if you do not already hold it, add from the last fill, take profit from average (default 3.14%). After each close, tax on net profit, then a self-dividend comes off the desk and the rest grows the next ticket. TP, tax, and self-dividend are knobs. A name 30% under the first fill drips 1% of money in it, once a month. One new name per bar — not per calendar day when you pick 4h or 1h.

Why 3.14% — is that because of π?

It is a named target, not a law of price. π is how you remember 3.14%. The only math that must hold is: 3.14% has to clear fees on the way in and out and still leave a sliver. If you treat π as a prophecy, you left the system.

Why DCA from the last fill, not the average?

Average entry lies when you have already added. Last fill is the print you actually paid. Five percent under that print is a new, worse price — not a cheaper story about the bag.

Does the next buy stay at $200?

No. The first ticket is capital ÷ slices. After a take-profit close, leftover after tax and self-dividend is added to the next ticket — until it hits the max first ticket (default $500). Default is reset: the cycle starts over at the original slice so cash remains for DCA. Switch to compound if you want the ticket to keep growing with no cap. Drip size is still 1% of that bag, not the rolling ticket. Optional overlays (off by default) can still shrink or grow that note: Vol Note from BTC vol, Drawdown Gear from desk drawdown.

What are Vol Note and Drawdown Gear?

Money overlays on Pi Slice, Band Fade, and OTT Fade. They do not change RSI, bands, the Harvest stack, or the take-profit. Pi Slice and OTT Fade default them off. Band Fade defaults them on. Vol Note sizes the ticket from BTC realized volatility versus a 50% annualized target (double vol → half note, clamped). Drawdown Gear cuts the same note from the desk’s own peak: full until −6%, half until −12%, quarter until −20%, then $0 until the drawdown is back under 6%. A crash that is only in one small ticket barely moves gear; a crash that hits the wallet de-levers.

How can the backtest show a loss if it paid self-dividend?

Self-dividend is paid from each closed TP winner, then removed from the desk. “Desk return” is only cash plus open lots still sitting there. Open bags are marked to the last close and never booked as a losing closed trade, so win rate can look high while the book is red. Add “Desk + taken” to see equity plus dividends (and tax) that already left.

What is Band Fade?

A 1h mean-reversion book on the same top-25 USDT list. Bollinger 20 with 3.5σ. Long when the hour trades 0.50% through the lower band; short when it trades 0.50% through the upper. Take-profit 1% from average. No stop. Max 5 names at a time. If a lot is underwater, last-fill DCA and the 30% bag drip are the same as Pi Slice. Vol Note and Drawdown Gear are on by default. Leverage is optional. After a winner you sit until the next stretch — you do not immediately re-enter the same name.

What is OTT Fade?

A counter-trend book on the Harvest stack (OTT 200 / 500 / 1000, VAR, 5% — same lines as OTT Harvest, not Co-Pilot 200/500/750). Clock is 3m, 5m, 15m, 30m, or 1h. Short when the stack is up and the bar tags 5% or more above OTT 1000. Long when the stack is down and the bar tags 5% or more below OTT 1000. Mixed stacks (adjacent lines closer than 0.15%) are skipped. Take-profit 0.75% from average. No stop. Top-50 book, max 10 names. Same Pi Slice ticket shop. Buying that 5% stretch on an up stack would be a chase — this desk fades it.

Education only — not financial advice. Strategies lose money. Paper is kinder than live. Not affiliated with Binance or TradingView.