The week’s tradable anomalies, the catalysts behind them, and how last week’s calls played out.

A token suddenly dominating crypto Twitter, whales dumping ETH, price spiking while network usage flatlines — you get the gist. We call these anomalies, and they often lead to trading opportunities.

This week we lead with the setups worth acting on, give you the backdrop they sit against, and grade how last week’s calls actually played out. Let’s dig in!

Santiment’s Weekly Anomaly Report will become its own newsletter soon — join the mailing list.

This Week in Brief

• Ripple, $XRP: three separate signals fired on the same day, including 25 times the normal amount of dormant coins moving, with the most negative sentiment of any asset we looked at.
• Polygon, $POL: eight days of increased attention while sentiment stayed negative and price sat still, then a +10% breakout on double the volume.
• OK, $OKB: up 25% in ten days, now fading — and the network behind it has about a hundred active addresses a day, pretty weak.
• Polkadot, $DOT: the week’s biggest attention spike landed at a record low, and leveraged buyers are still being wiped out.

New This Week

We added a ninth anomaly. ‘old_coins_moved’ flags an unusual amount of long-dormant supply moving, measured against each asset’s own history. It covers the 300 largest assets by market value and makes its first appearance below — including one reading at roughly 200 times the asset’s normal daily level. Read it as context on what holders are doing, not as a buy or sell signal.

Market Backdrop

Bitcoin stayed in its range, sliding to about $62.8K on August 16 before recovering to roughly $64.7K by August 18, up about +2% across the window. Ether did the same thing, going from $1,878 to $1,916. Bitcoin has now spent weeks between $60K and $67K with volatility at multi-year lows, and it sits roughly 49% below its October 2025 high near $126,198.

One signal woke up after three quiet weeks. ‘hyperliquid_avg_funding_rates’ fired once, reading -53.00 on August 15, against a trigger level of -20. Funding is the payment that flows between traders holding opposite sides of a perpetual futures bet. When it goes negative, the people betting against the price are paying the people betting bullish on it, which means the crowd is heavily positioned short. Three days later, price rose and $16.34M of positions were force-closed on August 18 — 96% of them shorts. The crowded bet got run over, which explains most of the week’s upward movement.

Large ETH holders have almost stopped selling. ‘eth_whale_dump’ fired once, at about $7.55M on August 18. Across three weeks the sequence now reads five events with a largest of roughly $46.6M, then five with a largest of about $18.7M, and now only one at $7.55M.

The macro backdrop is unusual. The Fed held rates at 3.50–3.75% at its July 28–29 meeting, but three regional presidents — Beth Hammack, Neel Kashkari, and Lorie Logan — dissented in favor of a rate increase, the first time three officials broke the same way since September 2016. Markets put the odds of a September hike near 36.6%. Minutes from that meeting were released on August 19, the day this window closed. Government bonds sold off worldwide overnight into that release, pushing yields to multi-year highs and pulling stocks down. Spot Bitcoin funds took in about $298M on August 18, ending a three-day run of outflows, though the past week was still about $385M negative. The Fear and Greed index is now sitting neutral, at the time of release of this report.

So the setups below sit in a downward and sideways trending market where the biggest single driver of the week was crowded traders being forced out of a bet.

This Week’s Trading Setups

1. Ripple ($XRP)

Signal: social_dominance_spike + project_in_trends + old_coins_moved · Lean: Volatility, non-directional

What’s anomalous. XRP is the only asset this week to trigger three different signals, and all three point at August 17.

‘project_in_trends’ fired twice, peaking at 293.64 on August 17 at 19:00 UTC — the highest trend score of the week, well above Bitcoin’s 143.61. ‘social_dominance_spike’ fired three times, peaking at 3.42 on August 17 at 12:00, with sentiment reading -1.85 on the spike day, the most negative of any asset we selected, easing to -0.85 since. And ‘old_coins_moved’ recorded 4.03 times its trigger level: about 1.1% of circulating supply moved, untouched for an average of 518.95 days.

Price ignored all of it. XRP went from $1.004 on August 12 to $0.992 on August 16 — a new cycle low of $0.9877 was reported that day — then back to $1.0014 by August 18, and only now beginning to pick up at $1.0686. Market cap is about $62.7B.

This stood out. On the day 25 times the usual dormant supply moved, XRP’s exchange volume was $928.9M, below its recent average of roughly $1.0B to $1.6B. If those old coins were being sold, the selling would show up as unusual exchange volume, and it did not. They likely moved between wallets.

Why. August 17 was the opening day of the Wyoming Blockchain Symposium in Jackson Hole, running August 17–20, co-hosted by SALT and Kraken. It is invitation-only with around 500 attendees, Ripple CEO Brad Garlinghouse is speaking, and SEC Chair Paul Atkins is attending. That accounts for the attention.

The dormant-coin movement has no confirmed explanation. One related event: on August 13, Ripple moved 50M XRP from an address linked to an operational wallet to an unidentified wallet. One million of those tokens went to a Binance-linked address the same day. The other 49M had not moved as of August 14. That sending address had received 150M XRP from Ripple’s main wallet on August 7.

Meanwhile the demand side got worse. Weekly net inflows into U.S. spot XRP funds collapsed 93%, from $14.86M to $1.01M for the week ending August 8 — the weakest week since those funds launched in late 2025, with net assets around $964M. XRP trades roughly 71% below its July 2025 high of $3.65.

What to watch. Track the 49M XRP still sitting in that unidentified wallet. If it moves toward an exchange, wallet housekeeping becomes real supply, and that may indicate incoming price suppression.

The takeaway: three signals, one day — when attention, dormant coins, and a scheduled event all land on the same date, the useful question is what you think whales will do. We think they will dump.

Trader tip: when 'old_coins_moved' fires, check that day's exchange volume before acting. Old coins moving is not old coins selling, and volume is how you tell them apart. It can still be the step before selling, so check where the coins went: wallet to wallet is housekeeping, onto an exchange is a seller getting ready.

2. Polygon ($POL)

Signal: social_dominance_spike · Lean: Long bias

What’s anomalous. ‘social_dominance_spike’ fired eight times, more than any other asset this week, peaking at 5.62 on August 15 at 12:00 UTC across a window running August 13 to August 17. Sentiment stayed negative the whole time: -0.13 on the spike day, -0.36 now. Market cap is about $878.45M.

For most of that stretch, nothing happened. Price sat between $0.0738 and $0.0748 from August 13 through August 16, and volume actually shrank, hitting its weekly low of $21.9M on August 16.

Then it moved. Price jumped to $0.0801 on August 17, up about +7% in a day, and reached $0.0818 by August 18 — roughly +10% off the August 16 close. Volume doubled to $70.1M on August 17 and stayed elevated at $63.5M. Futures volume rose about 334% to $109M with open interest up 10.37%, so traders added exposure into the move rather than just closing out old bets.

Why. There was no announcement on August 17. Price broke above a falling trendline that had capped it for weeks, and volume came with the break. Seems like a pure technical analysis play, though Polygon itself has some stuff going for it.

The groundwork was laid earlier in the month. Polygon’s Ithaca upgrade, its fourth mainnet hard fork in under five months, activated at block 50,185,000 on July 29 and added automatic failover so payments keep clearing when a block producer fails. The network passed 7.8 billion total transactions. Then on August 5, Polygon Labs restructured itself into a payments company, committing $250M and acquiring Coinme, alongside layoffs.

What to watch. The risk is not on the chart. POL trades about 94% below its peak, and the open question is whether the token captures any of the value from Polygon Labs’ payments business. A company doing well and its token doing well are two different outcomes.

The takeaway: the crowd stayed negative and was wrong.

Trader tip: negative sentiment plus lasting attention is only half a contrarian setup. The other half is volume confirming when price finally moves. Here it did, on the same day.

3. OK ($OKB)

Signal: price_network_activity_divergence · Lean: Short bias

What’s anomalous. ‘price_network_activity_divergence’ fires when price jumps more than 3% in a day without a matching rise in network usage. OKB read 107.54 on August 15. The only two other assets that fired read 0.13 and 0.06, so this was the one notable event.

The price move is real. OKB went from $85.88 on August 5 to $107.75 on August 14, about +25% in ten days, with $98.65 to $107.75 of that inside the window. Volume backed it up, running $15M to $32M a day normally and then hitting $75.9M on August 13, roughly 2.5 times normal. Market cap is about $2.13B.

It has been fading since. Price slipped to $99.09 by August 17 and sat at $101.11 on August 18, about 6% off the high. Volume drained back to $30.3M. Volume peaked on August 13 and price peaked on August 14, which is the ordinary order of events for a move that is ending rather than starting.

The caveat that makes this setup: OKB had between 40 and 125 active addresses a day this month. On August 14, its highest price of the month, there were 125. OKB is an exchange token: most of it sits inside OKX rather than moving between wallets, so its on-chain footprint is expectedly tiny. For an asset with about a hundred daily users, “price rose without network activity rising” is close to always true.

Why. No clear catalyst — flagging on data alone, so treat this as lower conviction. The standing context is that OKB’s supply was permanently fixed at 21 million tokens in the August 2025 burn, so there is no new issuance and there are no unlock dates. Intercontinental Exchange, the NYSE’s parent company, invested in OKX in March 2026 at a $25B valuation, with OKX users expected to be able to trade tokenized NYSE stocks from late 2026, and OKX has said it is weighing a U.S. listing. None of that is dated to this week.

What to watch. Whether volume returns. Reclaiming the $107.75 high would restore the trend. Continued lower highs on falling volume points back toward the $94 to $98 area the move started from. OKB is still roughly 58% below its August 2025 high near $258.

The takeaway: a divergence signal on an asset with almost no network — when the usage side of the comparison is near zero by design, the signal is only measuring the price move, so use it as a momentum flag and get your actual read from volume and the greater context of what’s happening.

Trader tip: before trading any divergence print, look up the asset’s daily active addresses and scrutinize the business context.

4. Polkadot ($DOT)

Signal: social_dominance_spike · Lean: Short bias

What’s anomalous. ‘social_dominance_spike’ fired four times, peaking at 10.33 on August 18 at 00:00 UTC. That is the highest peak of any asset we selected, and it was still firing when the window closed on August 19 at 02:00. Sentiment was negative throughout: -0.50 on the spike day, -0.45 now. Market cap is about $1.28B.

Price fell through all of it, from $0.773 on August 12 to $0.751 on August 18, about -2.9%, continuing a slide from $0.856 on August 4. DOT is down 11.5% over 30 days, 39.6% over 90 days, 58.7% this year, and roughly 98% below its November 2021 high of $54.87. It set a fresh all-time low near $0.75. Volume rose to $66.3M on August 18 as attention peaked, against $32M to $46M earlier in the week.

The crowd is still betting on a recovery. As of August 18, Binance’s account-based long/short ratio was 1.666 and OKX’s was 2.78, with Binance’s top trader accounts at 2.1888. Liquidations over that day ran about $1.57M, essentially all of it long positions against roughly $2,340 of shorts. Open interest sat at $174.68M, inside the range it has held since June and well below May’s $200M-plus.

Why. No fresh catalyst inside the window. The nearest dated event is Grayscale withdrawing its spot DOT fund application on August 7, days before a regulatory milestone. Beyond that, Polkadot has been losing activity to competing layer-1 networks all year, with active addresses trending down.

What to watch. Leveraged buyers are still being carried out, and that usually has to finish before a floor forms. DOT is in oversold territory inside a falling channel, which is a condition rather than a trigger. $0.76 is the immediate ceiling where the short-term averages meet. Losing $0.75 opens new lows with nothing underneath as reference.

The takeaway: negative sentiment with the crowd still long — attention arriving at a record low looks like a wave of people giving up, but many leveraged long positions still remain.

Trader tip: check the long/short ratio before trusting a sentiment reading — DOT's crowd sounds bearish while still holding longs at nearly 3-to-1 on OKX.

Last Week’s Scorecard

Scorecard read: One clean win, two partials, one miss. Every level we published held. CRV stopped at $0.239. BSV failed at $15.50. LINK never tested $8.17. But two of four directional calls did not pay, so the levels were doing more work than the directional predictions. We did miss an easy short bias prediction on $CRV given the move already happened and it was a common pattern that happened last year too. Need to be more thoughtful about setups like that, it was a good opportunity to make a short trade on fading good news.

What Else Triggered

The new signal’s largest reading came with a lesson attached. ‘old_coins_moved’ produced 10 hits across 9 underlying assets, and $FLOKI dwarfed everything else. A reading 200 times normal, for one day. Inside that number, 1.3% of circulating supply moved at an average dormancy of 480.33 days, and a single 50.00B transfer that had sat untouched for about 1,156 days — a little over three years — accounted for about 99.0% of the reading.

Price did nothing yet(opportunity?). FLOKI went from $0.00002025 on August 12 to $0.00002002 on August 18, down roughly 1%. The largest dormant-supply event of the week produced no price response at all, which is exactly what this signal’s documentation says to expect.

The duplicated Ethena, $ENA and staked-ENA contract series recorded 3.73 times its trigger level for ‘old_coins_moved’: about 5.4% of circulating supply moved, untouched for an average of 229.96 days. We count that as one asset rather than two.

The Sandbox, $SAND was the week’s one positive-sentiment spike** and went nowhere. Seven ‘social_dominance_spike’ triggers, peaking at 5.61 on August 15, with sentiment at +0.89 on the spike day easing to +0.17. Price moved from $0.0382 on August 12 to $0.0385 on August 18. Attention plus good feeling plus no price movement is the least useful combination we track, which is why it sits here rather than above.

Shiba Inu, $SHIB and Litecoin, $LTC each took a single ‘project_in_trends’ print** with no repetition — SHIB at 134.77 on August 13, LTC at 109.54 on August 17. Bitcoin, $BTC logged three ‘project_in_trends’ observations and a 143.61 peak, below XRP’s 293.64, and stays outside the non-BTC and non-ETH group by design.

No-shows this week: two of the nine tracked anomalies produced no eligible events — ‘social_price_correlation’ and ‘social_dev_score’. Correlation has now been silent across all seven assets it covers for four straight weeks.

Worth Watching

• XRP, $XRP: whether the 49M XRP still sitting in the wallet that received Ripple’s August 13 transfer moves toward an exchange, which would turn wallet activity into real selling pressure.
• Polygon, $POL: the open question is whether the token captures any of the value from Polygon Labs’ payments business. A company doing well and its token doing well are two different outcomes.
• Polkadot, $DOT: whether the long/short ratio finally comes down, since leveraged buyers are still being liquidated and that usually has to end before a low holds.
• OKB, $OKB: Whether volume returns. Continued lower highs on falling volume is bad.

Full Trigger Table

Between 2026-08-12 03:00 and 2026-08-19 03:00 UTC, 156 trigger observations fired across 7 of the 9 tracked anomaly types, spanning 80 signal-and-asset entries.

• Market-cap filter: latest_marketcap_usd > $50M.
• Old-coins-moved values are token-days, meaning tokens moved multiplied by days sitting still, so they should not be compared across assets.

Anomalies We Track

• Network Activity & Price Divergence Anomaly — detects price surges that are not accompanied by a corresponding rise in network activity (real growth). Triggered only if price growth exceeds 3% on the same day, ensuring the alert is relevant for significant bullish moves.
• Project in Trends Anomaly — detects when a specific crypto token gains significant traction in online discussions, placing it among the top 10 most-mentioned words in crypto-related conversations.
• Old Coins Moved Anomaly — detects unusually large amount of long-dormant supply moving, measured against each asset’s own recent history. Triggers above 1.5 times that baseline. A trigger shows old coins moved, not that anyone sold them.
• Large USDC-USDT Mint Anomaly — detects daily combined $1 billion or larger minting events of USDC and USDT stablecoins on the Ethereum blockchain.
• Social Price Correlation Anomaly — detects whether social sentiment and price are behaving in an unusually aligned or unusually decoupled way. Currently runs on 1h X.com social data for bitcoin, ethereum, solana, dogecoin, xrp, trx, and cardano.
• Social Dominance Spike Anomaly — detects abnormal spikes in social dominance. Alerts when values exceed a statistical significance threshold of 3 standard deviations over 30 days.
• Social-Dev Score Anomaly — detects unusual activity compared to the recent 30-day history of social media metrics and development activity. The score is weighted 60/40 for social and dev activity.
• Hyperliquid Average Funding Rates Anomaly — detects when rates exceed predefined high/low thresholds of +40% and -20%, measured hourly.
• ETH Whale Dump Anomaly — detects potential large-scale ETH dumps by monitoring whale wallet behaviors. Uses a 500 ETH threshold and a 30-day holding period to distinguish normal trading from potential dump scenarios.

Thanks for reading this week’s Santiment Anomaly Report. For real-time anomaly alerts and the underlying data, head to Sanbase — and since the report is becoming its own newsletter soon, join the mailing list to get it straight to your inbox.

See you next week.

Anomalously yours,
Santiment

Disclaimer: This report is for reference purposes only and is not intended as financial or investment advice. Always DYOR.