About the Pulse

ALGOZILLA PULSE — ABOUT

Fear & Greed vs. AlgoZilla Pulse

Tested across 25 simple strategies on BTC, side by side.

The Fear & Greed Index gives crypto a single daily number. AlgoZilla Pulse gives every tracked asset its own hourly score. Different tools, overlapping intent. This page explains how the Pulse works, how it compares to Fear & Greed on 25 honest backtests, and — most importantly — how (not) to use it.

We picked 25 simple, easy-to-explain rules. You could just as easily try a thousand more — these aren't the "best" strategies, they're a representative spread that shows what the Pulse can do well, what it does poorly, and where Fear & Greed has a structural advantage. Read it as a feel for the signal, not as a leaderboard.

INERT 0–35 CHARGING 35–49 AWAKENING 49–56 RAMPAGE 56–70 ATOMIC 70–100

What is the Pulse

Hourly, per-coin, OHLCV-derived

The Pulse is a regime-classification score in [0, 100] computed from price/volume signals only. No surveys, no Twitter sentiment, no external feeds. Five components combine into one score: volatility, volume, multi-horizon trend, Mayer multiple, and a Pu-241-decayed shock overlay. Updated every hour for every tracked asset — about 90 coins today.

Five regimes, asymmetric cutoffs

The cutoffs [35, 49, 56, 70] are calibrated to the actual time-distribution of crypto cycles, not equally spaced quantiles. The center sits slightly above 50 because crypto spends more time in bullish regimes than bearish — the cutoffs reflect that reality.

No BTC anchor

Each coin's Pulse is calibrated against its own history — not blended with BTC's. We tested three architectures (per-coin, 10% BTC anchor, 25% BTC anchor); per-coin standalone won by +0.044 to +0.069 mcap-Sharpe in our internal backtests.

Causal, no lookahead

Every Pulse value at time t uses only data from before t. Expanding-window percentile ranks, backward-looking returns, sign-aware shock decay — all backward-only by construction. Backtest results reproduce in live trading.

How the comparison was done

Fear & Greed publishes once per day. AlgoZilla Pulse runs every hour — that's 24 reaction-points per day instead of one. To honour Pulse's native resolution, both signals are evaluated on hourly resolution: the Pulse uses its raw 1-hour values, and each F&G value is repeated across the corresponding 24-hour day (it doesn't update intraday, so this preserves exactly what F&G actually offers).

The same trading rules feed both signals on the same BTC hourly price series. Costs are 0.30% per side, 0.60% round-trip — realistic retail fees including spread and slippage. Long-flat or fractional only, no leverage, no short selling.

This setup makes Pulse's hourly edge visible. Rules that need precise entry-timing or fast regime-detection benefit from 24× more decision points; F&G's daily updates mean its rules can only react with up to a one-day delay. That's an honest disadvantage for F&G in this hourly comparison — and it's also exactly why the Pulse exists. We benchmark against BTC Buy & Hold (~$9.2k → $82k over 8.5 years).

Pulse vs Fear & Greed — side by side

Both signals on BTC since 2018 (the Fear & Greed Index started Feb 2018). Pulse resampled to daily for fair comparison. Drag to zoom; double-click to reset. The right axis shows BTC price (log) for context.

How does each signal trade?

25 naive trading rules on both signals — same rules, different signal. Rows where a signal beats B&H are highlighted green. Strategies are sorted with the winners on top: rules that beat Buy & Hold on the Pulse come first (S1–S3), then rules that beat it on Fear & Greed (S4–S12), then near-zero comparators, then the rules that lose money — useful as cost-drag illustrations.

Strategy AlgoZilla Pulse Fear & Greed
ReturnSharpeα vs B&H ReturnSharpeα vs B&H
Loading…
BTC Buy & Hold (reference) —

Why we read the Pulse as confirmation, not contrarian

Classic technical indicators (RSI, Stochastic, Bollinger Bands) were designed for price: a price that drops to an "oversold" reading is statistically likely to bounce, so the rule is "buy oversold". That's mean-reversion, and it works on price because price has no inherent direction.

The Pulse is different. It's a sentiment-score. A high Pulse means trend, momentum, volume and structure are already aligned bullish. A low Pulse means they are aligned bearish. The score itself already tells you the direction. Applying classical "buy when oversold" logic to a sentiment-score is a category error: it's like buying because the alarm is ringing.

So all rules below treat the Pulse as a strength-confirmation: a higher reading is reason to be (or stay) long; a lower reading is reason to step aside. The Stochastic, RSI and BB variants we test all use the upper end of the band as the buy-signal, the opposite of how they would be applied to price.

What does each strategy actually do?

Plain-language description of every rule, what behavior it captures, and when it tends to work or fail.

S1 · Stochastic strength · %K > 80 Pulse top

Rule: %K over N=14 days on the Pulse. Long when %K > 80; flat when %K < 20.

Stochastic is normally used for mean-reversion on price; here we flip it because the Pulse is the sentiment, not a price proxy. The 14-day window plus the strict threshold dampen hourly flicker enough to clearly beat Buy & Hold even at 0.60% round-trip. The strictest member of the Stoch sweep (S5/S8/S7/S9/S6 are looser variants).

S2 · Pulse > 49 for 3 days best Sharpe

Rule: Long only when Pulse stays > 49 for three consecutive days (= 72 hourly bars); exit the moment any drops below.

Anti-flicker on the AWAKENING cutoff. The 3-day persistence requirement collapses fakeouts into a single transition. The clearest illustration that persistence is what makes Pulse rules survive realistic fees.

S3 · Pulse > 49 for 2 days Pulse winner

Rule: Same persistence idea as S2 but with a 48-hour confirmation window instead of 72.

Slightly more responsive than S2 — earlier entries, slightly more trades. Still beats B&H comfortably; confirms the persistence-window length is a tunable knob, not a binary.

S4 · EMA(90) crossover

Rule: Long while signal is above its own 90-day EMA; flat below.

Trend-following on a slowly-smoothed signal. Powerful on F&G (which already moves slowly anyway); on the noisier hourly Pulse this same rule fires too often and gets eaten by fees. A clear example of how update-frequency interacts with rule-design.

S5 · Stochastic strength · %K > 65

Rule: Looser version of S1 — entry threshold 65 instead of 80.

Earlier entries → more trades → more fee-drag on the Pulse. Works on F&G because daily updates already throttle turnover.

S6 · Long > 90-day rolling mean

Rule: Long when current signal exceeds its own 90-day rolling average; flat below.

A bias-filter cousin of S4. Same dynamic: F&G beats B&H, hourly Pulse flickers across the line too often.

S7 · Stochastic strength · %K > 70

Rule: Same as S1/S5 with entry threshold 70.

Middle of the Stoch sweep. F&G beats B&H, Pulse close to break-even but not over.

S8 · Stochastic strength · %K > 60

Rule: Loosest entry threshold of the sweep — 60.

Earliest entries; fires the most often. The "too aggressive" anchor of the sweep — useful to read alongside S1 to see how threshold-strictness translates into Pulse-side performance.

S9 · Stochastic strength · %K > 75

Rule: Penultimate step toward S1 — entry threshold 75.

Returns clearly positive, but slightly under Buy & Hold once Pulse fees apply. F&G version still beats B&H.

S10 · Long when Pulse > 53 (mid-AWAKENING)

Rule: Long whenever Pulse > 53 (= midpoint of the AWAKENING band); flat below.

Single-bar threshold rule. F&G barely beats B&H thanks to daily updates throttling turnover; on hourly Pulse the same rule loses badly to fee-drag (see S19/S20 for shorter-window variants of the same idea).

S11 · Long when Pulse > 49 (AWAKENING entry)

Rule: Long whenever Pulse > 49 (= the CHARGING→AWAKENING cutoff); flat below.

Simplest bias filter on the regime boundary. Same single-bar pattern as S10. The persistence-rules S2/S3 use the exact same threshold but require multi-day confirmation — and that's all the difference between green and red on the Pulse side.

S12 · Pulse > 53 for 3 days

Rule: Persistence variant of S10 — Pulse must stay > 53 for 3 consecutive days.

Slight loss on Pulse, slight win on F&G. Sits between the symmetric persistence winners (S2/S3) and the single-bar losers (S10/S11). Confirms persistence does most of the work.

S13 · Stochastic strength · %K > 90

Rule: Strictest of the Stoch sweep — entry threshold 90.

Above S1's threshold of 80. So strict that very few entries trigger; misses too many ride-ups. Marginally below B&H — useful to bracket S1 from above.

S14 · Long when Pulse > 56 (RAMPAGE entry)

Rule: Long whenever Pulse > 56 (= AWAKENING→RAMPAGE cutoff); flat below.

Stricter trend confirmation than S10/S11 but still single-bar. Better than the lower thresholds but still loses to fee-drag on hourly.

S15 · MACD on Pulse

Rule: MACD line on Pulse (EMA-fast − EMA-slow, 12/26/9 canonical). Long when MACD > signal line.

Classic trend-confirmation indicator. Survives but doesn't shine — the underlying EMA-cross still fires often enough on hourly Pulse that fees take a chunk.

S16 · ATOMIC-confirmed regime-flip

Rule: Stay long. Once ATOMIC has been touched, exit only after Pulse confirms a flip down (below AWAKENING). Re-enter on next AWAKENING upcross.

Don't pre-emptively trim — let the regime confirm. Returns clearly positive on both signals (Pulse +396%, F&G +669%) but doesn't quite beat Buy & Hold once realistic fees apply. The "good idea, just not quite enough" of the lineup.

S17 · Stochastic strength · %K > 85

Rule: Between S1 (80) and S13 (90).

Slight loss on both signals. Useful to read across the full S8 → S5 → S7 → S9 → S1 → S17 → S13 sweep — the curve is non-monotonic, peaking around %K>80.

S18 · Pulse > 49 for 24 hours

Rule: Persistence with a 1-day window — half of S3's window, a third of S2's.

Below B&H on both signals. Confirms persistence-window length matters: 1 day too short to filter out flicker fully, 2-3 days hits the sweet spot.

S19 · Pulse > 49 for 8 hours

Rule: Same persistence rule, 8-hour window only.

Cost-drag illustration. F&G isn't tested at this window because F&G updates only once per day (= 24 hours), so an 8-hour persistence rule is degenerate for it.

S20 · Pulse > 49 for 4 hours

Rule: Same persistence rule, 4-hour window — barely above the single-bar threshold.

The clearest cost-drag illustration in the lineup. F&G excluded for the same reason as S19 (degenerate vs daily updates).

S21 · Asymmetric: enter > 49 (3d) · exit > 70 (3d)

Rule: Long when Pulse stays > 49 (AWAKENING) for 3 consecutive days. Stay long through everything except a confirmed ATOMIC entry — only exit when Pulse stays > 70 for 3 consecutive days.

Easy to enter, hard to exit. Trades rarely, sits through pullbacks. A study in asymmetric thresholds. Doesn't outperform the symmetric persistence rules (S2/S3) here — the lack of a downside-exit hurts in deep crashes.

S22 · 3-state: enter > 35 (2d) · exit ATOMIC or INERT (2d)

Rule: Long when Pulse stays > 35 for 2 consecutive days (= CHARGING confirmed). Exit when Pulse stays > 70 for 2 days (= ATOMIC distribution) or < 35 for 2 days (= INERT capitulation). Three regimes, three decisions.

More respectful of the actual five-regime cutoffs than S21 — uses the CHARGING entry as the long-trigger. The downside-exit kicks in early enough to dampen drawdowns but also cuts winners that briefly dip into INERT. Net result on the Pulse: positive return but well under B&H. F&G excluded — its smoother daily readings rarely cross 35 or 70 in a 2-day stretch, which makes the rule degenerate for it.

S23 · Avoid lowest band

Rule: Long always except when in INERT.

Skip only true capitulation territory. Captures most of the bull-cycle but still exits at the wrong moment in deep crashes (Terra-LUNA, FTX). Structural baseline rather than a tradable system.

S24 · Avoid highest band

Rule: Long always except when in ATOMIC.

Skip distribution territory near cycle-tops. Avoids the largest drawdowns but misses the steepest melt-up phases that often happen inside ATOMIC.

S25 · Hold with ATOMIC trim

Rule: Always hold 100% except in ATOMIC, where size drops to 50%.

Stay structurally long but reduce size during the highest-tension band. Few trades, low cost-drag — the gentlest possible regime-overlay on Buy & Hold.

Equity curves

Click a strategy to see how each signal would have grown $1 over the period. Higher is better. The dashed line is BTC Buy & Hold. Drag to zoom on the chart, double-click to reset.

How to use the Pulse

✓ Do use it as

  • A regime gauge. "Is this asset in a low-energy band or a high-tension band right now?"
  • A position-sizing input. Trim positions when an asset enters ATOMIC, re-add on AWAKENING transition (see S25 hold-with-trim, S16 ATOMIC-confirmed flip).
  • A cross-asset scanner. Sort 90 coins by Pulse to find which assets are statistically extended versus accumulating.
  • A macro-context filter. The mcap-weighted market Pulse on /pulse/ is a single regime read for crypto.
  • One input among many. Building block, not a strategy.

✗ Do not use it as

  • A day-trading entry/exit signal. Naive regime-cross trading on hourly Pulse generates hundreds of round-trips; cost-drag wipes out signal value.
  • A capitulation-buying tool with leverage. Terra/LUNA, FTX showed an oversold Pulse can stay oversold for days.
  • A precise top-call. ATOMIC = "high-tension," not "imminent reversal." Cycle-tops in 2017 and 2021 spent weeks in ATOMIC before flipping.
  • A standalone trading strategy. Best naive Pulse rules (S1, S2, S3) beat Buy & Hold modestly after fees, but only when persistence or wide hysteresis prevents fee-eating flicker. Without that protection, the same Pulse data loses to fees on hourly. Pulse is a feature, not a system.
  • A prediction. Describes what the regime IS, not what it will become.

What the hourly numbers tell us

Three patterns jump out of the table when you read it carefully:

  • Single-bar threshold rules collapse. S10, S11, S14 (long when Pulse crosses a single level) lose 90%+ on the Pulse side with realistic 0.60% round-trip fees. Every flicker across the line is a paid round-trip; the cost-drag adds up to total drawdown over 8 years.
  • Anti-flicker / persistence rules survive. S1 (Stochastic %K>80, 14-day window) and S2/S3 (Pulse > 49 for 2–3 days) don't fire on every cross — they require sustained agreement. Same Pulse data, same fees, very different outcomes.
  • F&G's daily updates are an accidental advantage here. A signal that updates only once per day automatically transitions less often, which means lower turnover and lower fee-drag. F&G doesn't react faster — it just trades less because it can't.

The cost-drag arithmetic, made concrete. Take a single-threshold rule like "long when Pulse > 53". On hourly data the Pulse moves around its own threshold all the time: 52 → 54 → 52 → 55 → 52 over 4 bars triggers 2 entries + 2 exits. At 0.30% per side that's 1.2% paid in fees over 4 hours — for nothing. Multiply by 8 years of hourly flicker and the rule eats itself alive. Persistence rules (S2/S3, S12) require N consecutive bars above the threshold before they fire, which collapses that flicker into a single transition. Same signal, same fees, very different outcome.

The lesson: reaction-frequency only translates into edge when paired with smart trade-management. Naive "long whenever Pulse is high" rules at 0.60% round-trip eat themselves on hourly data. The Pulse's hourly resolution gives the AlgoZilla pipeline more decision-points to work with — but those points are useful only because the rest of the bundle decides when not to act on them.

Is the AlgoZilla Pulse a strategy indicator?

Short answer: not on its own. It's a regime indicator that informs strategies — and inside AlgoZilla it does exactly that, by setting per-regime entry and exit thresholds. Below the long answer.

Why "yes" you could call it strategy-relevant

  • It separates regimes cleanly. The 5-state classification (INERT → ATOMIC) carries real signal — several rules in the table above turn that signal into a backtest that beats Buy & Hold even after fees.
  • It's per-coin and hourly. Unlike single-number indices, every tracked asset has its own Pulse updating each hour, so cross-asset comparisons and intra-day decisions become possible.
  • It's transparent and reproducible. Pure OHLCV in, deterministic score out. No surveys, no APIs, no opaque sentiment models. You can verify it independently.

Why "no" — why a single indicator is not a strategy

  • One dimension is not enough. Markets have trend, momentum, volatility, liquidity, macro context and structure. A single 0–100 score collapses all of that into one number. Useful, but lossy.
  • No position-sizing or risk-management built in. A score doesn't tell you how much to allocate or when to take risk off — it just tells you the regime.
  • No cost-management built in. As the table above shows, naive rules lose to fees on hourly data. A real strategy has to decide when not to act, not just when to act.

How AlgoZilla actually uses it: dynamic per-regime thresholds

Internally, the AlgoZilla model uses the Pulse to set the entry and exit thresholds of the trading layer dynamically per regime. In INERT or CHARGING the model raises the bar for entries (skip noisy bottoms) and lowers the bar for exits (protect capital). In RAMPAGE it does the opposite — earlier entries, later exits — because the regime supports holding through pullbacks. In ATOMIC the model trims size and tightens exits because the cycle-top tail-risk is structurally elevated.

That's the practical role of the Pulse: not "the strategy", but the regime-aware switch that tells every other component (oracle scoring, smart-money signals, cycle anchors, risk-mit, exit ladders) which operating point to run at right now. The Pulse decides the playing field; the rest of the 7,700+ candidate-feature pool decides the moves on it.

How AlgoZilla uses the Pulse internally

One feature in a 7,700+ candidate-feature pool

Internally, the Pulse feeds the regime-aware classifier inside our trading pipeline (codename Pythia). One of 7,700+ candidate features alongside oracle-scoring, multi-horizon trend, smart-money signals, cycle anchors and risk-mitigation. Re-fit per coin every two weeks via walk-forward cross-validation. The bundle's mcap-weighted Sharpe sits several times higher than any single rule on this page.

Per-coin, hourly, transparent

Per-coin calibration so a coin in INERT while BTC is RAMPAGE is visible. Hourly resolution so the pipeline reacts within a bar instead of a day. Transparent from OHLCV so backtests reproduce in live trading. No Twitter, no surveys, no third-party APIs.

See the live Pulse: the market Pulse page shows the mcap-weighted aggregate and a per-asset grid; the BTC Pulse page drills into a single asset with score, regime, distance to nearest band-boundary, and an interactive 8-year history chart. Each tracked coin has its own page (e.g. ETH, XRP, SOL, …).

Takeaways

  • Both signals carry information. Several simple rules beat BTC Buy & Hold on either Pulse or Fear & Greed. Neither index dominates across all rule-types.
  • The Pulse rewards persistence and patience. The clear pattern across all 25 strategies: rules that require persistence or strict thresholds survive; single-bar threshold rules get eaten by hourly fee-drag. The three standalone Pulse rules that beat Buy & Hold after 0.60% round-trip fees are S1 (Stochastic %K>80) and S2/S3 (Pulse > 49 for 2–3 days).
  • The Pulse is per-coin and hourly. Where Fear & Greed gives crypto one daily number, the Pulse runs hourly across every tracked asset — different use-case: regime-filtering, position-sizing, cross-asset scanning.
  • The Pulse alone is a feature, not a strategy. AlgoZilla's actual trading models combine the Pulse with 7,700+ other candidate features and re-fit per coin every two weeks. The bundle's walk-forward Sharpe is several times higher than any single rule shown here.

Disclaimer

No investment advice. The information on this page does not constitute investment, financial, or trading advice and you should not treat it as such. AlgoZilla does not recommend that any cryptocurrency be bought, sold, or held. Backtests reflect historical hypothetical performance net of realistic transaction costs (0.60% per round-trip); they do not guarantee future results and do not include slippage in volatile market conditions, exchange downtime, or tax effects. Conduct your own due diligence and consult a qualified financial advisor before making any investment decisions.