Spike Family
One signal, two strategies. Algotoria Spike is the next-generation family, selective by design: it stays out of the market most of the time and deploys only on high-conviction moves. Algotoria Spike Diversified and Algotoria Spike Stable trade that same Algotoria Spike neural signal on up to 40 high-liquidity perpetual futures (the simulated record trades 7: Bitcoin, Ethereum and five major altcoins); they differ in the collateral behind it, and the difference shows in the time under water: in the simulation Spike Diversified spends less time under water, and less deeply, at a similar maximum drawdown and Calmar ratio. Simulated results from 1 January 2024 to 30 September 2026, shown gross or net of an illustrative performance fee, against Bitcoin, BITA Crypto 10, Gold and the S&P 500. Neither strategy has managed client money yet.
- Next-generation
- Selective by design: dormant most of the time, deployed only on high-conviction moves
- Highest expected risk-adjusted performance, not yet fully proven (simulated)
- Best diversification
- Limited capacity (about 2.5 million USD on current estimates)
Logarithmic y-axis: equal vertical distances are equal percentage moves. Daily closes, rebased to 0.00% on 1 January 2024.
Out of sample from 1 August 2026: the models were fixed before 1 July 2026 and every rule and parameter before 1 August 2026, so later months were never used to choose anything; earlier months are a walk-forward simulation whose settings were chosen on that same history.
| Parameter | Spike Diversified RECOMMENDED | Spike Stable |
|---|---|---|
| CollateralCollateralWhat the account holds between and beneath its trades. Both strategies take the same trades; the collateral is the only difference between them.Spike Stable holds dollar-pegged stablecoins only, so its value moves least. Spike Diversified adds gold and US-equity tokens during drawdowns, under a proprietary allocation rule: the collateral is then spread across more issuers and is not tied to the US dollar alone. | Broader diversification Yield-bearing stablecoinsStablecoins other than USDTTo open USDT-margined perpetual positions, the account may borrow USDT against stablecoins other than USDT, and the borrowed USDT accrues interest at the exchange’s rate, borne by the account. Yield-bearing stablecoins (RLUSD, OKUSD, USDG) typically earn more than this cost, offsetting it and adding to returns. USDT collateral carries no such cost., plus a gold and US-equity sleeve held only during drawdowns Gold and US-equity tokens beside the stablecoins reduce the dependence on any single token issuer and on the US dollar alone. | Lower volatility Equal thirds of RLUSD, OKUSD and USDG yield-bearing stablecoinsStablecoins other than USDTTo open USDT-margined perpetual positions, the account may borrow USDT against stablecoins other than USDT, and the borrowed USDT accrues interest at the exchange’s rate, borne by the account. Yield-bearing stablecoins (RLUSD, OKUSD, USDG) typically earn more than this cost, offsetting it and adding to returns. USDT collateral carries no such cost. Dollar-pegged collateral only, so the account carries no gold or equity price swings between trades. |
| Longest drawdown, daysLongest drawdownThe longest stretch, in calendar days, from the moment the simulated account first fell below its adjusted peak to the moment it regained it, measured on the five-minute equity path. A stretch still open on the as-of date is counted to that date. | 56 | 103 |
| Ulcer indexUlcer indexRoot-mean-square of the adjusted drawdown over every five-minute reading, in percentage points. Unlike maximum drawdown it counts both how deep and how long the account stays under water; lower is better. | 0.812 | 0.832 |
| Adjusted max drawdownAdjusted max drawdownAlgotoria’s adjusted definition tests both ends of a fall over 24 hours, on the five-minute equity path. A new high counts only at the lowest balance of the following 24 hours; a low counts at full depth only if the account is still at or below it 24 hours later.So neither a transient print nor a crash candle reversed within the day sets the figure. Figures are gross; net drawdowns are larger by construction. | −3.9% | −3.0% |
| Raw max drawdownRaw max drawdownThe deepest fall below the running peak, with no 24-hour test at either end: every five-minute low counts in full, including a crash candle that recovered within the day. | −10.2% | −10.2% |
| Simulated Calmar ratioCalmar ratioCAGR divided by the adjusted maximum drawdown over the same period. It measures return per unit of drawdown risk; higher is better. | 22.36 | 26.99 |
| Simulated gross CAGRCAGRCompound annual growth rate of the simulated account from 1 January 2024 to 30 September 2026, time-weighted in USDT and before any performance fee. | +87% | +82% |
| Positive monthsPositive monthsCalendar months that closed with a gain, out of all months in the simulation. | 27 / 33 | 26 / 33 |
| Risk tiers (max drawdown)Risk tiersThe three drawdown budgets an investor can select at onboarding. A budget is a monitored ceiling, not an automatic stop: at two-thirds of it the Investment Committee reviews the account, and at the full budget the client elects to continue, lower the budget or stop.The budget is not a guarantee and can be exceeded. | 15% / 25% / 35% | 10% / 20% / 30% |
| Explore Spike Diversified | Explore Spike Stable |
Signal generation
Both strategies are driven by one fully automated signal, Algotoria Spike, and take the same trades: only the collateral differs. It is built to stay out of the market most of the time and to act only on high-conviction moves.
Each instrument is analysed continuously, around the clock, by a proprietary neural network that reads horizons from order-flow detail up to multi-week context and reduces them to a single directional conviction score. Every instrument has its own direction model, deployed as a multi-seed ensemble so that no individual training run can drive a trade. A separate model forecasts near-term volatility; it never opens a position and feeds the capital-preservation overlay below.
Entry: a deliberately high bar
A position opens only when conviction clears a high entry threshold, fixed in research on strictly historical data and then frozen. In simulation the bar is cleared on roughly 64 days a year, typically on one or two instruments at a time. Size is proportional to conviction, within fixed per-instrument weights and the 3× gross exposure cap.
Crash-candle entry guard
At every completed five-minute candle the strategy takes the median return of the five altcoin legs over the last candle and over the last three. If the one-candle median is −5% or worse, or the three-candle median −8% or worse, no new altcoin entry is opened at the next boundary. The guard never closes a position and never touches Bitcoin or Ethereum. It is a protection, not a modelled edge: it costs about 13 percentage points of annual return against the single-candle rule it extends.
Exits
Every position carries a full exit stack from the moment it opens: a volatility-scaled stop-loss, a take-profit, a trailing profit lock, a hard time-stop and exit on signal reversal. Each is enforced in code. Holding periods are typically measured in hours.
Capital-preservation overlay
The overlay rests the book after an exceptionally strong day and resumes trading, at normal size, once the volatility model’s average forecast over the previous day rises above its own long-run median. In simulation it is load-bearing for the drawdown bar.
Traded instruments
The strategy may trade up to 40 USDT-margined perpetual futures, and the Investment Committee may widen the set at any time. The simulated record trades 7: Bitcoin, Ethereum and five major altcoins. Six legs are long/flat; Ethereum is the only leg permitted to go short.
Execution
Algotoria Spike trades as a taker, not maker-first. Its signals fall at the start of fast moves, when an order resting in the book is filled mainly by the trades that turn against it. Orders are price-protected and marketable. Stop-losses and take-profits stay inside the strategy; none rests at the exchange. Leverage cap 3×, selectable at 3×, 2× or 1×; about 64 deployment days and about 156 trades a year; positions typically last hours.
Shallow drawdowns are a design target, not a guarantee
Algotoria Spike Stable was designed to a 10% backtest drawdown bar on the five-minute adjusted path at its 3× gross exposure cap, and its investor drawdown budget is 30%. A replay cannot capture every trading cost or market regime.
Crypto is a leveraged risk-on asset
Despite ETF adoption, Bitcoin’s daily volatility is three to four times that of the S&P 500. The asset class remains regime-dependent.
A simulation is not a track record
The figures on this page are replays, not results earned by clients. Future results will differ, often materially. Read the full Risk Notice before allocating.