Smart Money Positioning API on Hyperliquid: Real Long/Short Ratios From the Top 100 Wallets

What Smart Money Positioning Actually Means

Most long/short ratio data you find on crypto dashboards is misleading. The numbers from CoinGlass and similar aggregators count every open position equally, which means a $50 retail short on BTC counts the same as a $50 million whale long. The headline ratio looks balanced but the actual capital direction is invisible.

What you really want to know as a trader is different. Where is informed capital positioned right now? When the top 100 traders by 30-day volume on Hyperliquid are net long $340 million versus $165 million short on BTC, that is a directional signal. When they flip to net short, that is a different signal. When the smart money sample on HYPE shows a 0.16x long-to-short ratio (extremely short-crowded), that tells you something the funding rate alone cannot capture.

Until today there was no clean way to access this data programmatically on Hyperliquid. You could watch individual whale wallets through Hyperdash. You could read the funding rate. You could guess from social media. But there was no aggregated, refreshed, queryable signal.

Today we shipped that endpoint live in production.

Introducing /api/v1/smart-money-positioning

The endpoint is called smart-money-positioning and it lives at https://www.buildix.trade/api/v1/smart-money-positioning. A simple GET request with your API key returns the aggregated long and short open interest for any Hyperliquid perpetual, computed from the actual on-chain positions of a curated sample of around 80-100 of the top traders by 30-day volume.

Sample call:

`bash curl -H "x-api-key: YOUR_KEY" \ "https://www.buildix.trade/api/v1/smart-money-positioning?symbol=BTC" `

Sample response:

`json { "symbol": "BTC", "exchange": "hl", "computed_at": 1777307422, "data_age_seconds": 204, "long_oi_usd": 342965548.59, "short_oi_usd": 165407031.84, "long_short_ratio": 2.0735, "wallets_sampled": 81, "wallets_long": 48, "wallets_short": 33, "coverage_pct": 23.4331, "market_oi_usd": 2169460480.56, "sample_quality": "strong" } `

Every field is meaningful. Long and short open interest are summed in USD across the sampled wallets at current mark prices. The ratio above 1.0 means the smart money is net long, below 1.0 means net short. Wallets long versus wallets short tells you whether the bias is concentrated in a few large positions or spread across many traders. Coverage percentage tells you what slice of the total market open interest the sample captures, so you can weight the signal accordingly per symbol.

The sample_quality field is derived server-side from coverage. Strong means the sample captures more than 20 percent of market OI (BTC, ETH). Fair means 10 to 20 percent. Thin means 5 to 10 percent. Symbols below 5 percent return a 404 with insufficient_coverage so you never get a noisy signal back.

Live Numbers Right Now

As of this article going live, here is what the smart money sample shows on the four major Hyperliquid perpetuals.

BTC ratio 2.07x with 24 percent coverage and strong sample quality. The top 81 wallets are carrying $343 million long against $165 million short. Forty-eight wallets long, thirty-three short. This is a clear directional bias and the kind of signal that historically precedes upward continuation.

ETH ratio 1.45x with 19 percent coverage and fair sample quality. Net long but less aggressively. Twenty-four wallets short are notable — when ETH smart money has more short wallets than long but the ratio is still net long in dollars, it usually means a few large traders are carrying conviction longs against a wider distrust crowd.

SOL ratio 1.39x with 7 percent coverage and thin sample quality. Direction is up but the sample captures less of the total SOL market because SOL has more retail dispersion. Use the signal with the caveat in mind.

HYPE ratio 0.16x with 7 percent coverage and thin sample quality. This is the most interesting reading of the four. A 0.16x ratio means the smart money sample is carrying six times more short exposure than long on HYPE right now. Even with thin coverage, the magnitude is the kind of asymmetry that a funding rate model alone cannot surface clearly.

Why This Matters For Bots and Quant Traders

If you run automated strategies on Hyperliquid, you have probably tried to incorporate positioning data before. The available options were limited. CoinGlass long/short ratio is broken because it weights every position equally and includes massive amounts of retail noise. Hyperdash leaderboards are useful for watching individual whales but expose no aggregate API. Funding rate is a proxy for crowding but lags actual position changes and gets distorted by funding-arb traders.

The smart-money-positioning endpoint gives you a different input. It is the literal aggregate of on-chain positions held by a sample of traders who consistently rank top by volume. That sample by definition skews toward informed flow because consistent high-volume traders survive only when they are net profitable.

How you use the signal depends on your strategy. A few patterns we have seen in our own backtests:

Trend confirmation. When your existing momentum or breakout signal fires, check the smart money ratio. If it agrees with your direction, the trade has a tailwind. If it disagrees, especially with sample_quality strong, you might be entering a contra-smart-money trade.

Mean reversion fading. Extreme ratios (above 2.0x or below 0.5x with strong sample) historically mark short-term exhaustion zones. When the sample is loaded one direction and price has already moved that way, the marginal entry has degraded.

Regime detection. Track the rate of change in the long/short ratio across BTC and ETH. When the ratio starts to drift directionally over multiple refresh cycles, that often precedes regime shifts in broader crypto markets by hours or days.

Cross-asset divergence. When BTC smart money is loading long but ETH is flat or short, that asymmetry often resolves through ETH catching up to BTC, not BTC retracing to ETH. The endpoint makes this kind of multi-symbol analysis trivial.

How the Sample Works

Transparency on what we actually do.

A curated pool of around 80-100 Hyperliquid wallets is maintained in our database. The pool is composed of three sources. First, traders who rank consistently in the top by 30-day volume on Hyperdash and similar public leaderboards. Second, wallets identified through our own orderflow analysis as having outsized impact on price discovery in specific markets. Third, a small set of manually curated addresses with public reputation as informed traders.

Every wallet in the pool is filtered through two checks. The wallet must have at least $200,000 in current open notional positions on Hyperliquid (active risk capital, not equity sitting in vaults). The wallet must have between 1 and 200 fills in the last 48 hours (active trader, not a flat zombie wallet, not a high-frequency market-making bot).

A cron job runs every 15 minutes that pulls the live position state of every active wallet through the standard Hyperliquid clearinghouseState endpoint, then aggregates by symbol with a SQL transformer. The result is upserted to our database and exposed through the API. Symbols where the aggregated open interest exceeds 200 percent of total market OI (a sign of double-counting or data error) are excluded as a sanity check. Symbols with coverage below 5 percent return 404 because the sample is too thin to be informative.

The whole pipeline runs on public on-chain data. We pay zero data fees, we expose zero proprietary information, and the methodology is auditable end-to-end.

What This Endpoint Does NOT Do

To set expectations clearly.

The endpoint does not predict price. It tells you where smart money is positioned right now, which is correlated with future returns but not deterministic. Use it as one input among many.

The endpoint does not cover HIP-3 builder-deployed perpetuals on Trade.XYZ, Felix, Kinetiq, Ventuals, and the other emerging dexes. The whale-monitor pipeline only fetches mainnet positions today. HIP-3 coverage is on the roadmap.

The endpoint does not capture retail flow. The sample is intentionally biased toward high-volume traders. If you want broad market sentiment including retail, use the funding rate or our /api/v1/liquidation-map endpoint which captures asymmetric exposure across the full market through a different model.

The sample size for any individual symbol is bounded. Coverage on BTC and ETH is meaningful (19-24 percent). Coverage on smaller caps is by definition thinner. We expose the coverage_pct field exactly so you can weight the signal appropriately.

Pricing

The endpoint is gated to the Whale tier at $79 per month. Whale tier also unlocks the full liquidation-map API, the deep-view orderflow analytics on every supported pair, AI Strategy Advisor with bring-your-own-key on six providers (OpenAI, Anthropic, Google, Groq, Mistral, Ollama), and priority webhook delivery on alerts.

We chose to gate this specific endpoint at the Whale tier because the engineering cost (whale-monitor cron, position aggregation, sanity-checking, schema upkeep) is non-trivial and because the use case (programmatic integration into automated strategies) skews toward serious users. Free, Starter, and Trader tiers continue to access the full screener, deep view orderflow, and signal alerts on more than 530 perpetual pairs.

There is no contract. You can cancel any time. The 24-hour Starter trial is available if you want to feel out the platform before going to Whale.

Getting an API Key

Sign up at buildix.trade, upgrade to Whale, and your API key appears in the dashboard within seconds. No KYC, no waiting list, no enterprise sales call.

The endpoint is documented at buildix.trade/docs along with the rest of the v1 API. Rate limit is generous (designed for cron-style polling at 1-minute intervals or faster). Cache-Control headers tell your client to cache for 60 seconds, which is more than enough given the 15-minute refresh cadence on the underlying data.

Why We Built This Now

Hyperliquid has changed what is possible for retail-accessible market intelligence. Five years ago, the kind of data this endpoint exposes was either nonexistent or locked behind $50,000 per month Bloomberg terminals. The on-chain transparency of Hyperliquid means a small engineering team can build it for under $1000 per month in infrastructure costs and offer it at $79 per month with healthy margins.

We are also seeing demand from a specific kind of user. Algorithmic traders who run their own bots, do their own backtesting, and need clean inputs they can trust. Those users do not want a black-box dashboard that tells them what to trade. They want raw data with clear methodology that they can integrate into their own systems. That is exactly what this endpoint is.

The other reason is timing. Funding rates have become noisier as more funding-arb capital has entered Hyperliquid. The signal-to-noise ratio of pure funding-based crowding indicators has degraded. Aggregating actual positions from the top traders side-steps that noise because actual positions are slower to flip than funding rates and are biased toward informed flow.

Try It Now

If you already have a Whale subscription, hit the endpoint right now and see what BTC, ETH, SOL, and HYPE smart money positioning looks like. If you do not, the upgrade path is two clicks from the dashboard.

`bash curl -H "x-api-key: YOUR_KEY" \ "https://www.buildix.trade/api/v1/smart-money-positioning?symbol=BTC" `

We will continue to expand coverage of the wallet pool, add HIP-3 support when whale-monitor reaches builder-deployed perps, and improve the sample quality classification as more historical data accumulates. Feedback from active users always shapes what gets prioritized next. If you build something interesting on top of the endpoint, we want to hear about it.

Start at buildix.trade.

FAQ

How is this different from Hyperdash leaderboards? Hyperdash exposes individual wallet positions through its UI but does not offer a public aggregation API. Our endpoint sums positions across a curated sample of top wallets and returns the aggregate ratio along with diagnostic metadata, designed for programmatic consumption.

Why is HYPE coverage only 7 percent? Smaller market cap perps have wider position dispersion across more retail wallets, so the top 80-100 traders capture a smaller share of the total HYPE open interest. The signal still surfaces directional asymmetry (the 0.16x ratio is informative even at thin coverage) but you should weight it against the coverage_pct as exposed in every response.

How often does the data refresh? Every 15 minutes via a cron job that polls the Hyperliquid clearinghouseState endpoint for every active wallet in our pool. The data_age_seconds field in every response tells you exactly how fresh the snapshot is, and the endpoint returns 503 data_stale if the cron has not refreshed within 30 minutes.

What happens if a symbol returns 404? Either the symbol does not exist on Hyperliquid mainnet, or the smart money sample has aggregated less than 5 percent of total market OI for it. The 5 percent floor is a sanity check to prevent noisy signals on long-tail markets where one or two wallets would dominate the result.

Can I use this for HIP-3 markets like xyz:BRENTOIL or vntl:OPENAI? Not yet. The endpoint returns 400 unsupported_market for any symbol containing a colon (HIP-3 compound symbols). We are working on extending whale-monitor to fetch positions across builder-deployed dexes in a future release.

Will the wallet pool change over time? Yes. We rotate the sample as wallets become inactive or as new top traders emerge from the Hyperliquid leaderboard. The methodology stays consistent (notional > $200k, fills 1-200 in 48h) so the signal remains comparable across time even as the underlying addresses turn over.

Is the methodology audited? The pipeline is internally documented and the source data is publicly verifiable on-chain through Hyperliquid's open API. We are open to discussing the implementation with serious users who need confidence in what they are integrating.