A study, not a pitch

Hold captures trends.
Perpeto works with volatility.

A 4.5-year study of three crypto assets, two strategies, and the one thing they don't share: how capital behaves between the peaks.

This is not a pitch. It is a backtest.

Between January 2022 and July 2026, we ran the Perpeto system on historical data for BTC/USDT, ETH/USDT and SOL/USDT, and compared it to a simple buy-and-hold investor. Same start date, same exchange data. One assumption: the investor had $2,000 to deploy on each asset on January 1, 2022. Everything else is what the market and the rules produced.

What follows is the full record — including where Hold wins.

Period
4.5 yrs
01.01.2022 → 06.07.2026
Assets tested
BTC · ETH · SOL
vs. USDT, Binance spot
Binance dataset
130M+
tick rows per asset, per year
Closed trades
12,461
BTC 2,209 · ETH 4,077 · SOL 6,175
§ 01 — Methodology

How we tested

Before the numbers, the rules that produced them.

The backtest runs on 1-second tick data from the Binance spot market for BTC/USDT, ETH/USDT and SOL/USDT — approximately 130+ million rows per asset, per year. We do not aggregate to daily candles for the simulation itself; the system sees price exactly as it would in production. Daily numbers in the charts below are roll-ups of that second-by-second engine, not its input.

The strategy ran with the identical rules that operate in production — no leverage, no margin, spot only, long only, progressive entries on drops, positions closed only in profit after a price reversal (callback). BTC and ETH used the Dynamic profile; SOL used the Standard profile — a less aggressive variant chosen for SOL's higher volatility.

Every position is the same size within each asset. This is not a martingale. There is no doubling, no revenge sizing, no discretionary override.

The benchmark is a passive buy-and-hold investor who deployed the same $2,000 on day one and did nothing else for 4.5 years. Same starting capital across all three assets for direct comparison.

Starting capital
$2,000 per asset (BTC, ETH, SOL)
Time horizon
1,648 days · 4.5 years
Data source
Binance spot · 1-second granularity (130+ million rows per asset, per year)
Rule set
Dynamic profile (BTC, ETH) · Standard profile (SOL)
Leverage
None · spot long only
Position sizing
$25 BTC · $20 ETH · $14 SOL (equal-weight within asset)
Exit rule
Profit-only close after callback reversal
What this study deliberately excludes
  • This is not a fit-to-history optimisation. The rules are production rules, unchanged.
  • Profit numbers below are net of Binance spot trading fees (already deducted from every closed position).
  • Does not include the Perpeto performance fee. The fee is tiered from 5% to 25% depending on your lifetime cumulative profit — the more you earn, the less you pay. Details at perpeto.com/#pricing.
  • Realised profit was not reinvested. The system kept the original position size for the full 4.5 years.
  • Does not model slippage during extreme volatility.
  • Past results are not a forecast of future returns.
§ 02 — Headline results

Same rules. Same capital. Three very different markets.

The system did not change. The markets did.

* Values as of 06.07.2026 Reinvested

Standard view is the primary study: realised profit is withdrawn as earned and position size stays fixed. Switch to Reinvested to see how the same $2,000 — with earned profit compounded back into larger positions, no new client capital — would have performed. The tables below and the charts in §03 update; the surrounding commentary describes the standard scenario.

Case A — The uptrend market

BTC / USDT

+34.2% price move · Hold's natural terrain

Between Jan 2022 and Jul 2026, Bitcoin moved from $47,723 to $64,042 — a 34.2% net gain with a 66% interim drawdown and a $124k peak in October 2025. This is the environment where Hold plays its home game.

Metric
Hold
Perpeto
Total P&L ($)Total profit or loss in dollars = final account value minus the $2,000 invested; includes both realised profit and the unrealised move of still-open positions.
+$684
+$810
Invested own capitalThe capital the client put in from their own pocket; in the standard scenario it is also the ceiling on deployed capital — with reinvestment, anything above it comes only from earned profit.
$2,000
$2,000
Total P&L (%)Total P&L divided by the $2,000 invested — how much the account earned relative to what the client put in (e.g. BTC: $810 / $2,000 = 40.5%).
+34.2%
+40.5%
Average yearly P&L (%)Total P&L (%) expressed as an average annual rate with compounding (same methodology as CAGR).
+6.7%
+7.8%
Average active capital ($)The average amount actually deployed in open positions over 4.5 years; the rest of the $2,000 sat as available exchange cash.
$2,000
$686
Average yearly return on active capital (%)Annual return on only the capital actually at work — P&L divided by average active capital, expressed per year (CAGR methodology).
+6.7%
+18.9%
Max drawdownThe largest peak-to-trough decline over the period — how deep the account fell at its worst moment.
−66.1%
−29.1%
Annualised volatilityHow much the equity fluctuated day-to-day, scaled to a yearly figure. Lower means a calmer ride.
50.6%
37.6%
Sharpe (rf=0)Return per unit of risk. Higher = more efficient conversion of volatility into return. Risk-free rate set to zero.
0.38
0.38
Closed tradesThe number of positions the bot opened and closed over the period — every one closed in profit.
0
2,209
Realised net profitProfit from closed trades after Binance fees, excluding the unrealised move of still-open positions.
$0
$1,012.79

Over the full period, Perpeto delivered a higher total return with less than half the drawdown. But at BTC's October 2025 peak, Hold was ahead — briefly worth $5,224 (+161%), while Perpeto was well behind at that same moment. Selling at that peak was not possible in practice; by July 2026, Hold had given back nearly half. Perpeto had not tried to catch the top, so it had less to give back.

Case B — The declining market

ETH / USDT

−52.2% price move · Hold's stress test

Between Jan 2022 and Jul 2026, Ethereum fell from $3,766 to $1,800 — losing 52% of its price. This is the environment where Hold has nothing to capture.

Metric
Hold
Perpeto
Total P&L ($)Total profit or loss in dollars = final account value minus the $2,000 invested; includes both realised profit and the unrealised move of still-open positions.
−$1,044
+$868
Invested own capitalThe capital the client put in from their own pocket; in the standard scenario it is also the ceiling on deployed capital — with reinvestment, anything above it comes only from earned profit.
$2,000
$2,000
Total P&L (%)Total P&L divided by the $2,000 invested — how much the account earned relative to what the client put in (e.g. BTC: $810 / $2,000 = 40.5%).
−52.2%
+43.4%
Average yearly P&L (%)Total P&L (%) expressed as an average annual rate with compounding (same methodology as CAGR).
−15.1%
+8.3%
Average active capital ($)The average amount actually deployed in open positions over 4.5 years; the rest of the $2,000 sat as available exchange cash.
$2,000
$855
Average yearly return on active capital (%)Annual return on only the capital actually at work — P&L divided by average active capital, expressed per year (CAGR methodology).
−15.1%
+16.8%
Max drawdownThe largest peak-to-trough decline over the period — how deep the account fell at its worst moment.
−74.0%
−35.3%
Annualised volatilityHow much the equity fluctuated day-to-day, scaled to a yearly figure. Lower means a calmer ride.
69.4%
35.4%
Sharpe (rf=0)Return per unit of risk. Higher = more efficient conversion of volatility into return. Risk-free rate set to zero.
0.11
0.40
Closed tradesThe number of positions the bot opened and closed over the period — every one closed in profit.
0
4,077
Realised net profitProfit from closed trades after Binance fees, excluding the unrealised move of still-open positions.
$0
$1,385.90

ETH was a declining market throughout most of the period. A Hold investor ended with 48% of their capital. Perpeto ended well above its starting capital. The difference is not timing luck — it is what the system did between the peaks: 4,077 closed positions, each in profit, funded by the same volatility that punished Hold.

Case C — The deep-crash market

SOL / USDT

−54.2% price move · the deepest stress test

Solana lost 95% of its value at the deepest point of the 2022 crash — falling from $179 to $8 before recovering. By July 2026 it closed at $82, still 54% below its starting price. The deepest stress test of the three assets.

Metric
Hold
Perpeto
Total P&L ($)Total profit or loss in dollars = final account value minus the $2,000 invested; includes both realised profit and the unrealised move of still-open positions.
−$1,084
+$1,083
Invested own capitalThe capital the client put in from their own pocket; in the standard scenario it is also the ceiling on deployed capital — with reinvestment, anything above it comes only from earned profit.
$2,000
$2,000
Total P&L (%)Total P&L divided by the $2,000 invested — how much the account earned relative to what the client put in (e.g. BTC: $810 / $2,000 = 40.5%).
−54.2%
+54.1%
Average yearly P&L (%)Total P&L (%) expressed as an average annual rate with compounding (same methodology as CAGR).
−15.9%
+10.1%
Average active capital ($)The average amount actually deployed in open positions over 4.5 years; the rest of the $2,000 sat as available exchange cash.
$2,000
$764
Average yearly return on active capital (%)Annual return on only the capital actually at work — P&L divided by average active capital, expressed per year (CAGR methodology).
−15.9%
+21.6%
Max drawdownThe largest peak-to-trough decline over the period — how deep the account fell at its worst moment.
−94.6%
−44.5%
Annualised volatilityHow much the equity fluctuated day-to-day, scaled to a yearly figure. Lower means a calmer ride.
94.6%
27.7%
Sharpe (rf=0)Return per unit of risk. Higher = more efficient conversion of volatility into return. Risk-free rate set to zero.
0.29
0.48
Closed tradesThe number of positions the bot opened and closed over the period — every one closed in profit.
0
6,175
Realised net profitProfit from closed trades after Binance fees, excluding the unrealised move of still-open positions.
$0
$1,485.00

SOL was the deepest test. Hold lost 95% of its capital at the worst point and recovered to just 46% by the end. Perpeto finished well above its starting capital, with a far shallower drawdown than Hold. The 6,175 closed positions during the crash and the subsequent volatility did what Hold could not — turn a falling chart into accumulated cash.

Hold gave back most of its peak. Perpeto kept it.

Reinvested

The system does not predict market tops — it just refuses to give them back. Peak equity for each strategy vs. final equity on 06.07.2026, on the same data, same exchange.

Strategy
Peak equity
Final equity
Loss from peak
BTC Hold
$5,224 06.10.2025
$2,684
−48.6%
BTC Perpeto
$2,925 26.10.2025
$2,810
−3.9%
ETH Hold
$2,566 22.08.2025
$956
−62.7%
ETH Perpeto
$3,198 27.10.2025
$2,868
−10.3%
SOL Hold
$2,927 18.01.2025
$916
−68.7%
SOL Perpeto
$3,335 03.10.2025
$3,083
−7.6%

In this backtest, when the market retraced, Hold gave back 49 to 69 percent of its peak profits. Perpeto gave back only a small fraction of its own. Same market, same time — different mechanism.

§ 03 — The process, visualised

What the system did, month after month.

The outputs above are consequences. These are the mechanics that produced them.

Equity curves — Hold vs. Perpeto
Reinvested
Fig. 01

Both strategies start at $2,000 on Jan 1, 2022. Hold's line is a 1:1 replica of the underlying price movement — the same shape scaled to the account's USD value. The gap between Hold and Perpeto is what the system does between the peaks.

Active capital & open positions
Reinvested
Fig. 02

All three assets start from the same $2,000 client capital ceiling. Position size scales to the ladder depth — $25 on BTC (80 max), $20 on ETH (100 max), $14 on SOL (143 max). The system expands into drawdowns and contracts on recoveries, holding a fraction of the ceiling in open positions on average and keeping the rest as exchange cash, available at any time. (Toggle Reinvested above to see how deployed capital grows when realised profit is compounded.)

Total equity vs. realised profit over time
Reinvested
Fig. 03

Every dollar of the user's account, visualised as two layers — both held on the user's own exchange. The upper area is total account equity, which moves with the price of the underlying asset. The lower area is realised profit accumulated, which only grows. Notice how realised profit climbs day after day regardless of whether the price is rising or falling — that floor cannot be undone by the next drawdown.

Monthly realised profit
Reinvested
Fig. 04

Closed, realised USD profit per month, net of Binance fees. Profit stays on the user's own exchange account — they can withdraw it at any time or leave it to compound through redeployment.

Currently showing: BTC/USDT
Activity
48.5%

of days closed at least one position. Every closed position was in profit. No stops, no overrides.

Adaptive exposure
80

maximum open positions during the deepest drawdown. System sized itself to the market.

Closed positions
2,209

total closed positions on BTC. Each one closed in profit.

Capital efficiency
$686

average active capital on BTC over 4.5 years; peak $2,000. The rest sat as available exchange cash.

§ 04 — Where the system is quiet about itself

What Perpeto did not do.

A backtest is worth the caveats it publishes alongside it.

01

Perpeto's peak was lower than Hold's peak at the top of the bull run.

At Bitcoin's all-time high on 6 October 2025, a Hold investor was momentarily sitting on +161% ($5,224 from $2,000). Perpeto was near +43% at the same moment. If you could have sold at that exact top — which no one does in practice — Hold would have delivered nearly 4× the Perpeto return in that instant. By 6 July 2026, however, Hold had given back nearly half of that peak. Perpeto had not tried to catch the top, so it had less to give back.

02

Mean reversion is an implicit assumption.

Every open position waits for price to recover above entry. In a sustained, multi-year decline without interim bounces, capital stays allocated. The system does not time bottoms — it waits for them. This works because crypto markets oscillate; it does not work if an asset only goes down.

03

Capital must be reserved.

Perpeto assumes the user keeps enough reserve on the exchange to fund positions during drawdowns. The backtest ran with disciplined capital management; a real user who under-funds the strategy will see different results. This is not a "set and forget" product for undersized accounts.

04

Profit numbers above are net of Binance fees only.

Binance spot trading fees are already deducted from every closed position. The Perpeto performance fee is not included. The Perpeto fee is tiered from 5% to 25% depending on your lifetime cumulative realised profit — the more you earn, the less you pay. See perpeto.com/#pricing for the full breakdown.

If after reading this section the system still makes sense to you, it probably does.

§ 05 — Conclusion

Hold captures trends. Perpeto works with volatility.

Hold

A position on time. It assumes the asset will be worth more in the future than it is today, and it trades patience for return. It is a good tool when trends are real and long.

Perpeto

A position on volatility. It assumes the asset will move up and down between now and the future — and it converts each oscillation into a closed, realised trade. It is a good tool when markets chop, drift, or decline without direction.

Between Jan 2022 and Jul 2026, Perpeto beat Hold on all three assets — BTC (+34.2% vs +40.4%), ETH (−52.2% vs +43.4%), SOL (−54.2% vs +54.1%). Three very different markets, one system, same shape of result. The profit Hold showed at its peak existed only on paper — and by the end of the period, most of it was gone. Perpeto retained 90 to 96 percent of its own peak.

Your funds stay on your exchange. Read + Trade API only. No withdrawal access.
Two versions of the same system

Without reinvestment, realised profit can be withdrawn at any time and each position stays the same size. With reinvestment, earned profit stays in and progressively enlarges each new position — the client never adds new capital. Same $2,000 start, same trades, same exchange data. Only the profit already produced is put back to work. The reinvesting version is available via the Standard / Reinvested switch in §02 — its effect on realised profit over 4.5 years:

BTC / USDT
$1,013$1,337
+32% realised profit
same $2,000 client capital
ETH / USDT
$1,386$2,045
+48% realised profit
same $2,000 client capital
SOL / USDT
$1,485$2,257
+52% realised profit
same $2,000 client capital

Reinvestment raises deployed capital above $2,000 only by putting already-earned profit back to work — never new client money. It also carries slightly more open profit into a retrace, so peak-to-final give-back is a little larger. Both trade-offs are visible in the toggled tables and charts above.