Classical technical structure computed on this coin: moving-average relationships, Bollinger squeezes, where price sits in its recent range, and RSI on both the daily and hourly.

Range position shows where today's price sits between its recent low and high. A squeeze means the price has gone unusually quiet, which often comes just before a big move — though it does not tell you which direction. RSI shows whether a move has run hot or cold; overbought means stretched, not that it is about to turn around.

Everything here is a standard textbook signal. Golden cross and death cross come from the moving-average relationship, Bollinger squeeze breakouts come from the volatility compression, and RSI divergence — price making a new high while RSI does not — is one of the oldest reversal warnings there is. Range position is the basis of simple mean reversion back to the 20-period average.

Design a new technical strategy for HEMI using /api/v1/indicators/technical/HEMI. Combine at least two of the structure signals — moving averages, Bollinger squeeze, range position, RSI — into one rule set rather than trading any single one. Write out entry, exit, stop and sizing. Validate on /api/v1/backtesting/klines with fees, and report which of the component signals actually contributed and which you could drop.

Every technical reading is archived per symbol in the daily bundle: the technical_regime block of /backtesting/daily-snapshots/{date}. If you want a different lookback than the one we publish, rebuild the indicators yourself from /backtesting/klines — minute bars for Hyperliquid perps and 450+ Binance markets.

Using CryptoDataAPI, backtest the classic technical signals on HEMI. Pull price bars from /api/v1/backtesting/klines and compute a 50/200 moving-average cross, a Bollinger squeeze breakout, and RSI divergence. Test each separately with taker fees and a fixed stop. Report win rate, profit factor and max drawdown per signal, and tell me which ones fail to beat buy-and-hold.

A three-state trend classifier, red / grey / green, built on ADX and DMI with hysteresis so it does not flip back and forth on noise the way a raw crossover does.

Three colours, and that is the whole indicator. Green means a clear uptrend, red a clear downtrend, grey means no real trend and probably not worth trading. It waits for solid evidence before changing colour so it does not flicker back and forth. Days since the last flip matters: a colour that has only just changed is the least settled.

ADX and DMI, which drive this indicator, are most often used as a trend FILTER rather than an entry: run moving-average crossover or breakout systems only when the colour is red or green, and stand aside in grey. That single filter is what keeps trend-following systems from bleeding out during sideways markets.

Design a new trend strategy for HEMI around the SIGNUM state at /api/v1/indicators/signum-rgg/HEMI. Treat the red/grey/green colour as the filter and invent the entry trigger yourself, including what you do during grey. Define stops and how you re-enter after a flip. Validate against the signum_rgg block of /api/v1/backtesting/daily-snapshots/{date} plus /api/v1/backtesting/klines, and report how much of the return came from simply avoiding chop.

The signum_rgg block inside /backtesting/daily-snapshots/{date} holds the recorded red/grey/green state per symbol per day, so you can test the colour changes without recomputing ADX. Raw price history for your own version is at /backtesting/klines.

Using CryptoDataAPI, test the SIGNUM red/grey/green state as a trend FILTER on HEMI. Pull the signum_rgg block from /api/v1/backtesting/daily-snapshots/{date} across the archive and price bars from /api/v1/backtesting/klines. Run a moving-average crossover strategy twice: once unfiltered, once trading only while the state is red or green. Report how much the filter improved profit factor and drawdown.

The model's probability distribution for the next 24 hours: direction, volatility, liquidation risk, funding direction, open interest, and which regime it is most likely to move to.

These bars are odds, not promises. If high volatility reads 38%, it means that in past situations that looked like this one, big price swings followed about 38% of the time. Nothing here says what WILL happen — only what usually happened next. When the bars are all roughly even, the model genuinely does not know.

Probabilities are what position-sizing rules need. Fractional Kelly sizes a bet from its odds and payoff rather than from conviction, and expected-value filters skip trades whose odds do not justify the risk. Traders also use the volatility bucket to set stop distance, so a stop is wide enough to survive normal noise in the current conditions.

Design a new position-sizing strategy for HEMI driven by the 24h probability distribution at /api/v1/quant/coins/HEMI?horizon=24h. The entry signal can be simple; the point is that size scales with the odds and the volatility bucket. Define the sizing formula explicitly and cap the worst case. Validate on /api/v1/quant/history plus /api/v1/backtesting/klines, and report the drawdown against a fixed-size version of the same entries.

/quant/history returns the stored model features and outputs per timestamp, which is what you need to check whether the probabilities were actually calibrated: bucket past readings and compare the predicted rate against what really happened next.

Using CryptoDataAPI, check whether the 24h probabilities for HEMI are calibrated. Pull /api/v1/quant/history, bucket the predicted probabilities into deciles, and for each bucket compare the predicted rate against what actually happened over the next 24 hours. Plot predicted versus realised and report the calibration error. Say plainly if the model is over-confident.

Dealer gamma exposure derived from perp positioning: where market makers are forced to buy into strength and sell into weakness, or the reverse, plus the gamma flip level where that behaviour inverts.

Market makers take the other side of everyone's trades, then buy or sell to protect themselves. Above the flip price, that protective trading pushes AGAINST moves, so the price tends to stay in a range. Below it, their trading pushes the SAME way as the move, so a fall can snowball. The flip level is simply the price where that behaviour switches over. The percentage next to "GAMMA FLIP LEVEL" (e.g. -3.67% FROM MARK) is how far the CURRENT price sits from that level — not a forecast of where price is headed. Negative means the flip is that far BELOW the current price: you are on the calmer, range-bound side right now, and price would need to fall roughly that much to cross into the accelerant zone. Positive means the flip is ABOVE the current price: you are already on the accelerant side, and price would need to rally roughly that much to reach the calmer zone above it. The closer the number is to zero, the closer price is sitting to that switchover line.

Gamma scalping is the strategy this measures: dealers buy dips and sell rallies to stay hedged, which pins price near the flip level. Traders fade the extremes while the market sits above the flip, and switch to breakout tactics below it, where the same hedging accelerates moves instead of damping them. The flip level itself is a common support/resistance marker.

Design a new trading strategy for HEMI built on dealer gamma. Read the live signal from /api/v1/quant/gex?symbol=HEMI. Use the flip level and the amplify/dampen state to define entries, exits and stop placement — the two states should trade differently. State your assumption about dealer hedging, then validate on /api/v1/backtesting/snapshots?data_type=gamma_exposure and report whether the flip level actually mattered.

Gamma exposure is archived as its own snapshot type: /backtesting/snapshots?data_type=gamma_exposure, first written 2026-07-06. Each record carries the dealer positioning and the gamma-flip level, so you can test how price behaved on either side of the flip.

Using CryptoDataAPI, test the gamma-flip level on HEMI. Pull /api/v1/backtesting/snapshots?data_type=gamma_exposure and price bars from /api/v1/backtesting/klines. Compare realised volatility and mean-reversion in the hours when price sat ABOVE the flip against the hours BELOW it. Report whether fading moves above the flip and following them below it would have beaten doing the same thing regardless of the flip.

The largest open perp positions in this coin on Hyperliquid: how big they are, which way they lean, and how that has shifted. Hyperliquid settles on-chain, so these are real positions rather than survey estimates.

These are the biggest individual bets on this coin. Watch the changes more than the totals: almost everyone being positioned for a rise is normal while prices are rising. What matters is when the big players suddenly switch sides or cut their size. Being big does not mean being right — large traders get forced out too.

This is the raw material for copy trading and smart-money following, where you mirror the largest verified positions. The contrarian version is the crowded-trade fade: when the biggest players are all one way, their exit is the fuel for the move against them. Pair it with liquidation-cluster hunting to find the prices where that exit gets forced.

Design a new trading strategy for HEMI built on whale positioning from /api/v1/quant/whales. Decide whether you are following the big positions or fading them, and justify which from the data rather than from intuition. Define entry, exit and how you handle a whale being liquidated. Validate on /api/v1/quant/whales/history?days=180 plus /api/v1/backtesting/klines, and report the hit rate for both directions so the choice is evidenced.

/quant/whales/history?days=180 gives the position history directly. For the underlying per-trader rows use /backtesting/snapshots?data_type=hl_trader_positions, which is what the roll-up is built from and lets you follow individual wallets across time.

Using CryptoDataAPI, test whether following the largest HEMI positions works. Pull /api/v1/quant/whales/history?days=180 and price bars from /api/v1/backtesting/klines. Build a signal from the net whale long/short skew, test both following it and fading it, and report forward returns at 4h, 24h and 7d for each. Include how often the whales were simply wrong.

Scheduled and unscheduled catalysts for this coin: news scored for real price impact, security incidents, and upcoming token unlocks that add supply.

Things that could move the price: news, security problems, and token unlocks — dates when a batch of new coins is released to early investors and team members, adding to the supply. An empty panel genuinely means nothing is scheduled, rather than data failing to load.

This feeds event-driven and catalyst trading. Token unlocks support a supply-shock trade — the size and date are public, so the weakness often arrives before the unlock rather than on it. Scheduled news drives the buy-the-rumour, sell-the-news pattern, where the move happens into the announcement and reverses on it.

Design a new event-driven strategy for HEMI using /api/v1/event/regime/HEMI and the unlock schedule. The rules must be executable in advance — no reacting to news after the fact — so define how many days before a catalyst you act and what closes the position. Validate on /api/v1/backtesting/news-events and /api/v1/backtesting/snapshots?data_type=token_unlocks, and be explicit about any lookahead you had to avoid.

/backtesting/news-events is the archived catalyst tape with the same scoring you see here. Token unlocks have their own snapshot type, /backtesting/snapshots?data_type=token_unlocks, which matters because unlock dates are known ahead of time and are therefore testable without lookahead.

Using CryptoDataAPI, test how HEMI trades around scheduled catalysts. Pull /api/v1/backtesting/news-events and /api/v1/backtesting/snapshots?data_type=token_unlocks, then measure returns from 7 days before each unlock to 7 days after, using price bars from /api/v1/backtesting/klines. Report whether weakness arrived before or on the date, and whether unlock size relative to float predicted the size of the move.