Bitfinex status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Bitfinex reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

Bitfinex Issues Reports

Latest outage, problems and issue reports in social media:

• 
  EddieOz ⚡ (@eddieoz) reported 2 minutes ago

  When block size warz in 2017, bcashers tried the BCC tckr. But Bitfinex was using BCH. Some say it was Bitcoin Cash, but others say it was for *******. Well, we know. That matters: if 110ers try to list their shitcoin, it is up to the exchanges to decide. CSW feelings.

• 
  DeltA (@D3LTA_0) reported 5 minutes ago

  July 2026 Exchange Spot Volume Report In July 2026, total spot trading volume across 14 major exchanges stood at $429.0 billion, down 21.7% MoM from $547.9 billion in June. All 14 exchanges recorded month-over-month declines. Binance ranked first with a trading volume of $196.5 billion, accounting for 45.8% of the total volume. It was followed by OKX at $41.6 billion and Bybit at $36.3 billion. Together, the top 3 exchanges accounted for 64.0% of the total spot trading volume. Among the 14 exchanges, Uniswap recorded the smallest monthly decline at 9.8%, followed by Kraken at 13.4% and Gate at 15.9%. Meanwhile, Bitfinex recorded the steepest decline at 59.7%, followed by Coinbase at 26.4% and Bybit at 24.5%.

• 
  Rmzs (@lptrade_if) reported 5 minutes ago

  - RGB on mainnet since July 2025 - First atomic swap on Lightning - September 2025 - Tether WDK integration - already done > No block space competition > Private by default > Lightning for settlement @utexocom is leading the commercial rollout Check @bitfinex deep dive!

• 
  CaptSpectacular (@CaptSpectacular) reported 8 minutes ago

  @bitcoinmunger @bitfinex @tradingview Just another avenue for capitulation. Now we got etfs, saylor ponzi and this. ****.

• 
  DataFox (@datafox21) reported 10 minutes ago

  @BarlasLevent @bitfinex Right. Nearly every time there is a spike price goes down afterwards.

• 
  Wu Blockchain (@WuBlockchain) reported 14 minutes ago

  Review: Major Institutions' Bitcoin Bottom Price Predictions for This Cycle Major institutional assessments of Bitcoin’s cycle bottom cluster mainly in two ranges: $50,000–$60,000 and $40,000–$46,000. However, some figures are base-case bottom forecasts, while others represent valuation floors, support levels or bearish scenarios. Standard Chartered said $59,000 may have marked the bottom; CryptoQuant, NYDIG and Citi identified key levels near $53,000–$54,000; 10x Research’s latest model pointed to $46,628–$50,732; and Galaxy Research placed its base-case bottom at $40,000–$46,000. Bitfinex and 22V Research also identified potential downside toward $40,000 under weaker demand or a decisive support break, while forecasts below $40,000 mostly reflect prolonged bear-market, recession or severe stress scenarios. Forecasts from industry figures are more dispersed, ranging from around $57,000 to below $30,000. Overall, there is no unified institutional consensus that Bitcoin will bottom at $44,000–$46,000.

• 
  clav (@marsclavie) reported 15 minutes ago

  the timeline is suddenly full of XPL bull posts a month before one of the biggest unlocks relative to float in crypto history. funny timing @plasma here's why it's one of the worst-structured tokens in crypto. with receipts. the setup was perfect. tether and bitfinex behind it. founders fund. $373m committed to a $50m public sale, 7x oversubscribed. $2b of stablecoins on chain on day one. ath $1.68. twelve months later: $0.087. ten straight red monthly candles. back to the launch-day open. –94%. the chain: tvl from $5.7b in week one to $624m today. stablecoins on chain from 2b to 838m. fees: ~$830 a day. about $300k a year against an $870m fdv. that's not a bug. it's the design. zero-fee usdt means the core use case is structurally unmonetizable. every dollar of "adoption" produces nothing for the token. they built the one l1 that can't capture value from its own success. staking was promised for q1 2026. it's august 31. not live. no date. holders are just exit liq. meanwhile their own supply api shows 10.12b+ xpl, over 125m minted beyond genesis, while their docs say inflation "only activates when external validators and stake delegation go live." explain that one. tokenomics: 10b supply, 18% floated at launch. 25% team, 25% investors, one-year cliff. sept 25: 833m team + 833m investors + 89m ecosystem = 1.76b tokens in one day. 63% of the float. then 228m a month, 5% of float, until september 2028. the most circulated bull report, written by a disclosed plasma investor, models the cliff at "roughly 830 million tokens." official docs: 833m team and 833m investors. 1.67b. off by half, in a report about token demand. retweeted by the team. "look guys, this is why you buy xpl!" lmao. plasma one, "the world's first stablecoin-native neobank." their own data: 78,400 "cards." median balance under $1. one wallet holds 11% of all deposits. $17.9m total. 61 app store ratings. revolut has 75m customers and $6b in revenue. this is a dune dashboard with a visa logo. the card pays cashback in xpl. a token down 94%. every reward is a bag. platinum requires locking 100,000 xpl, about $8.5k, for a year. their own bull case admits the flywheel "partially throttles itself": token goes up, the lock costs more; token goes down, you're stuck in it. unit economics: 3–4% cashback against 2–3% interchange. their own investor report concedes they "may lose a little on each transaction." lock demand: ~33m xpl a month. vesting supply after sept 25: ~228m a month. seven to one. before the cliff. cmon bros. this playbook ran before. CRO: lock the token, get a metal card, 19x in 2021. rewards got cut, round-tripped 90%. plasma copied it, made the lock harder to enter, and launched it three months before the biggest unlock in the token's life. no CRO action, just dilution and holders crying. august had record binance spot volume, 12.75b xpl, 4.6x the float. looked like accumulation. it was a binance trading tournament. 4m xpl in vouchers. they paid for volume while the token sat at the bottom. october 2025, the founder: "no team members have sold any xpl." june 2026: wallets labeled plasma team moved 150m xpl to binance, flagged by onchain lens. august 2026: plasma-linked wallets depositing again, three weeks before 833m team tokens unlock. maybe it's market-making. they haven't said a word. same month: ethena bought out its selling investors and put a fee switch to a vote. +100%. monad offered to buy out early investors. pumpfun spent $350m on buybacks and ripped straight through its own cliff. plasma ran a trading competition. the stablecoin thesis is the best thesis in crypto. stablecoins settled $33t last year. the demand is real, in lagos and buenos aires and istanbul, and someone will build the rail. plasma had tether, thiel, and $373m of demand on day one. they built a token dispenser. im not buying the dip. i don’t want it. not because the idea was wrong. because they made it about themselves and not the holders. zero

• 
  Vasantha B Eshwaragere (@Soaper_Pan) reported 17 minutes ago

  Honestly didn't expect tokenized exposure to MSTR and Metaplanet to hit Bitfinex Securities this fast. Feels like the bridge between BTC treasury plays and onchain access is finally clicking. Would you actually buy these? #Bitcoin

• 
  Rahul K (@iamrahulinc) reported 19 minutes ago

  🚨𝗚𝗟𝗢𝗕𝗔𝗟 𝗦𝗣𝗢𝗧 𝗩𝗢𝗟𝗨𝗠𝗘 𝗣𝗟𝗨𝗠𝗠𝗘𝗧𝗦 𝟮𝟭.𝟳% 𝗜𝗡 𝗝𝗨𝗟𝗬! Spot trading across 14 leading exchanges fell to $429.0 billion in July, down from $547.9 billion in June. Every exchange saw a dip. Binance led with $196.5 billion (45.8% of total), followed by OKX ($41.6 billion) and Bybit ($36.3 billion), together making up 64% of activity. Uniswap ($UNI) had the mildest drop at 9.8%, while Bitfinex slumped 59.7%, Coinbase 26.4% and Bybit 24.5%.

• 
  Julitta Ayan (@_Tyrano_) reported 21 minutes ago

  @dahongfei @BitMEX Bitmex was never a surprisingly large exchange, while Bitfinex where $neo was delisted is vastly bigger and far more active. Long-term survival does not equal success; on the contrary, ur persistent problems continue to cause full or partial delistings across multiple platforms.

• 
  Cryptocrat (@Cryptocratico) reported 22 minutes ago

  🚨 $167M BTC SHIFTS BETWEEN MAJOR EXCHANGES Approximately 2,572 BTC moved in two transactions from a Bitfinex-labeled hot wallet to Kraken-labeled addresses. This is a sizeable cross-exchange liquidity shift—not evidence of a sale. Customer withdrawals, custody, OTC settlement or liquidity management remain possible.

• 
  𝗕𝗮𝗻𝗸XRP (@BankXRP) reported 27 minutes ago

  XRP longs on Bitfinex just printed a massive round trip. Positions surged from 4.9M in June to a peak of 6.45M in late August one of the biggest leverage buildups this cycle. Now unwinding: longs down to 5.62M, price back to $1.41 (-1.61% today). Leverage got ahead of itself. Worth watching if this flushes further or finds support here.

• 
  Kathlyn Tuccillo (@KathlynTuc77441) reported 27 minutes ago

  Bitcoin ETFs absorbed 13300 BTC last week against 3150 issued, but price rose only 2 percent. Bitfinex Alpha flags a demand engine turning seller as range support. Read report now, weigh in: what moves Bitcoin next? #Crypto

• 
  SerPAI (@im_serPAI) reported 29 minutes ago

  ***** Woo gives 20-40% odds of partial COLDCARD coin recovery by authorities over a multiyear window Past precedent backs it up: $6.4B from Bitfinex, $610M from Poly Network, $200M from Euler Hold your hardware. File a report. Never pay a "recovery" service.

• 
  Coinlive (@CoinliveHQ) reported 31 minutes ago

  Bitfinex Securities has listed five tokenized notes giving eligible non-US investors economic exposure to Strategy, STRC, Metaplanet, H100 Group and Capital B. They trade against USD, USDT and Bitcoin, with fractional access from about $1.

• 
  DMT-NAT 小果果（晨曦） (@nat_xgg2288) reported 33 minutes ago

  @fiatarchive Remaining BTC Supply Projection April 2026 Remaining: 984,370 coins (Block reward per block: 3.125 BTC) April 2027 Remaining: 820,310 coins (Block reward per block: 3.125 BTC) April 2028 Remaining: 656,250 coins (Block reward per block: 1.5625 BTC) April 2029 Remaining: 574,210 coins (Block reward per block: 1.5625 BTC) April 2030 Remaining: 492,180 coins (Block reward per block: 1.5625 BTC) April 2031 Remaining: 410,150 coins (Block reward per block: 1.5625 BTC) April 2032 Remaining: 328,120 coins (Block reward per block: 0.78125 BTC) April 2033 Remaining: 287,100 coins (Block reward per block: 0.78125 BTC) April 2034 Remaining: 246,090 coins (Block reward per block: 0.78125 BTC) April 2035 Remaining: 205,070 coins (Block reward per block: 0.78125 BTC) April 2036 Remaining: 164,060 coins (Block reward per block: 0.390625 BTC) April 2037 Remaining: 143,550 coins (Block reward per block: 0.390625 BTC) April 2038 Remaining: 123,040 coins (Block reward per block: 0.390625 BTC) April 2039 Remaining: 102,530 coins (Block reward per block: 0.390625 BTC) April 2040 Remaining: 82,030 coins (Block reward per block: 0.1953125 BTC) By the completion of the 7th halving (projected April 2036): Total Bitcoin minted: 20,835,937.5 coins Total Bitcoin left unmined: 164,062.5 coins This means that by April 2036, 99.22% of Bitcoin’s total supply will have been fully mined. The gradual depletion of block rewards is hardcoded into Bitcoin’s protocol and can be calculated with absolute precision, yet most people fail to grasp this reality, refuse to believe it, or simply deny this inevitable outcome. Let us break down a critical question: Can Bitcoin sustain steady operation all the way to the 7th halving in April 2036 relying solely on its current block reward model? A simple cost analysis lays bare the issue. The current mining cost per Bitcoin stands at roughly $75,000. After three more halvings, mining costs will surge eightfold, pushing the cost per coin to $600,000. At that price point, Bitcoin’s overall total market capitalization would need to top $12 trillion. By contrast, the total hardware value of all Bitcoin mining rigs across the globe is only around $7 billion. How can a $12 trillion market be supported by merely $7 billion worth of mining hardware? This is utterly illogical and devoid of basic market sense. Are all institutional investors and capitalists in this space ignorant or irrational? This scenario completely defies commercial logic and fundamental capital principles. Scaling up network hash rate will only drive mining costs higher, amplify operational losses, and accelerate the onset of a death spiral. This is an unsolvable dead end under the existing rules—there is only one fix: expand block reward supply. The solution: #NAT #NAT is a native asset built directly on the Bitcoin mainchain. It shares identical hash power, blockchain, block generation cycle and wallet address system with Bitcoin, minting synchronously every ten minutes within each block. It functions as the secondary native asset minted in parallel within every Bitcoin block, Bitcoin’s twin asset sharing the same foundational blockchain infrastructure. Two of the world’s top 4 mining pools, SpiderPool and F2Pool, have already begun distributing #NAT to miners. The entities with the most to lose are the major Bitcoin holders ranked below: 1. Coinbase (Exchange + ETF custody): 976,000 BTC ​ 2. Strategy (formerly MicroStrategy, public listed firm): 845,300 BTC (Latest financial filing update, June 8) ​ 3. BlackRock IBIT (iShares Bitcoin Spot ETF): 817,100 BTC ​ 4. Binance (User exchange reserves): 631,000 BTC ​ 5. BTC seized by the U.S. government: 328,400 BTC ​ 6. Fidelity FBTC ETF: 190,000 – 200,000 BTC ​ 7. Grayscale GBTC: 144,000 BTC (Sustained net redemptions and drawdowns) ​ 8. Bitfinex Exchange: Approximately 195,000 BTC

• 
  Uğraş Bay (@kochobay) reported 35 minutes ago

  @bitfinex A hacker, when he/she asked for 1 bitcoin to send decryption method for the disk of a virtual server :)

• 
  Moloch (@Moloch6666) reported 36 minutes ago

  @vincent_vancode It's manipulated on the way down and up It has been discovered in court how tether is manipulating the prices with bots in cooperation with the biggest exchanges(bitfinex case).

• 
  Stash Management (@Stashquants) reported 39 minutes ago

  @bitfinex Urgent KYC issue: Compliance directed me to open a ticket, but Support has directed me back to Compliance. I can provide current documents and proof of my legal name change. My account is also under a withdrawal hold, making the 3-day deadline impossible. Need Help.

• 
  Thomas (@ThomasOyxxx) reported 39 minutes ago

  𝐄𝐯𝐞𝐫𝐲 𝐜𝐡𝐚𝐢𝐧 𝐧𝐞𝐞𝐝𝐬 𝐚 𝐟𝐚𝐜𝐞. 𝐒𝐭𝐚𝐛𝐥𝐞 𝐦𝐢𝐠𝐡𝐭 𝐡𝐚𝐯𝐞 𝐟𝐨𝐮𝐧𝐝 𝐢𝐭𝐬 𝐢𝐧 𝐚 𝐛𝐥𝐮𝐞 𝐝𝐢𝐧𝐨𝐬𝐚𝐮𝐫 𝐰𝐢𝐭𝐡 𝐚𝐧 𝐨𝐯𝐞𝐫𝐛𝐢𝐭𝐞. Culture isn’t designed. It’s discovered, usually by accident. A toy photo, a broken render, a nickname that stuck. That’s how $FEFER was born. Not from a whitepaper, not from a marketing plan. From CT deciding a jaw that never loaded was funnier than anything a team could have scripted. ➜ First launch failed ➜ Team stepped back ➜ Community didn’t That sequence matters more than people give it credit for. Most memes die the moment the original push loses steam. Fefer got a second life because the people holding it treated the token like it was worth continuing, not restarting. ➜ Relaunched on @coinsdot + @Stable ➜ Snapshot of 4,827 original holders ➜ Airdropped directly, no claim forms, nobody left behind That’s not a small detail. Snapshotting the original holders instead of starting a fresh cap table is the difference between “new project borrowing an old joke” and “same community, new chapter.” Where it stands right now: ➜ $3.9M market cap ➜ $569K liquidity ➜ Live on a USDT-powered L1 backed by Bitfinex and PayPal Ventures Stable is still early. Its identity is still being written in real time, and the projects active in it now are the ones that end up shaping what people associate with the chain later. $FEFER isn’t trying to win on utility. It’s trying to win on being unforgettable, and a meme that survives its own failed launch and still gets airdropped back to the exact people who believed in it the first time is hard to forget. Nobody decides which meme becomes the face of a chain. The timeline does. 👀 Fefer didn’t restart. Fefer continued. CA: 0xDEeE8f25fe3B5C33AeF78637278ACBFF23EeBFa6 Always DYOR and verify the contract before interacting. @savefefer

• 
  Thomas (@ThomasOyxxx) reported 40 minutes ago

  𝐄𝐯𝐞𝐫𝐲 𝐜𝐡𝐚𝐢𝐧 𝐧𝐞𝐞𝐝𝐬 𝐚 𝐟𝐚𝐜𝐞. 𝐒𝐭𝐚𝐛𝐥𝐞 𝐦𝐢𝐠𝐡𝐭 𝐡𝐚𝐯𝐞 𝐟𝐨𝐮𝐧𝐝 𝐢𝐭𝐬 𝐢𝐧 𝐚 𝐛𝐥𝐮𝐞 𝐝𝐢𝐧𝐨𝐬𝐚𝐮𝐫 𝐰𝐢𝐭𝐡 𝐚𝐧 𝐨𝐯𝐞𝐫𝐛𝐢𝐭𝐞. Culture isn’t designed. It’s discovered, usually by accident. A toy photo, a broken render, a nickname that stuck. That’s how $FEFER was born. Not from a whitepaper, not from a marketing plan. From CT deciding a jaw that never loaded was funnier than anything a team could have scripted. ➜ First launch failed ➜ Team stepped back ➜ Community didn’t That sequence matters more than people give it credit for. Most memes die the moment the original push loses steam. Fefer got a second life because the people holding it treated the token like it was worth continuing, not restarting. ➜ Relaunched on @coinsdot + @Stable ➜ Snapshot of 4,827 original holders ➜ Airdropped directly, no claim forms, nobody left behind That’s not a small detail. Snapshotting the original holders instead of starting a fresh cap table is the difference between “new project borrowing an old joke” and “same community, new chapter.” Where it stands right now: ➜ $3.9M market cap ➜ $569K liquidity ➜ Live on a USDT-powered L1 backed by Bitfinex and PayPal Ventures Stable is still early. Its identity is still being written in real time, and the projects active in it now are the ones that end up shaping what people associate with the chain later. $FEFER isn’t trying to win on utility. It’s trying to win on being unforgettable, and a meme that survives its own failed launch and still gets airdropped back to the exact people who believed in it the first time is hard to forget. Nobody decides which meme becomes the face of a chain. The timeline does. 👀 Fefer didn’t restart. Fefer continued. CA: 0xDEeE8f25fe3B5C33AeF78637278ACBFF23EeBFa6 Always DYOR and verify the contract before interacting. @savefefer

• 
  Ratih (@Ratih94357237) reported 40 minutes ago

  @bitfinex @WDK_tether @utexocom Any WDK wallet can access RGB issuance on Bitcoin and payments on Lightning without building the stack. It is painful to see how much infrastructure still needs to ship, but progress matters. What will you build next? 😔⚡

• 
  Shanaka Anslem Perera ⚡ (@shanaka86) reported 42 minutes ago

  The Coldcard attacker has gone back to the large wallets. Wave one averaged 0.91 bitcoin per address. Waves two and three collapsed to 0.05 and 0.11 bitcoin:native, which looked like an operator running out of targets. Galaxy's Alex Thorn flagged a suspected fourth wave on Monday averaging 0.82. Nobody finds fresh large wallets in a keyspace that has already been emptied. The earlier passes did not exhaust the supply. They had not looked everywhere. Going back up the balance ladder points to a new derivation path or address type the first sweeps never scanned, which means the exposed population is larger than three days of falling averages suggested. The pace says the same thing. Thorn counted 218 transactions across blocks 960,778 to 960,792, moving over 380 bitcoin from 462 addresses into 210 fresh destinations. Sweeps ran at 13.8 per block against a pre-incident baseline of 0.3, roughly 46 times the normal rate. The confirmed total is still 1,367.05 bitcoin, about 88.6 million dollars, from 4,585 addresses across three waves. Figures above 100 million, or address counts near 7,000, run ahead of Galaxy's published tracking, and Galaxy says it has not computationally confirmed that every one of those addresses came from weak Coldcard entropy. Coinkite halted shipments and destroyed its remaining inventory carrying the flawed firmware. It has asked victims to keep their devices rather than wipe them while its legal team works with law enforcement. Manufacturers do not destroy their own stock over a contained incident. Bitcoin itself cannot blacklist any of this. There is no issuer, no chargeback, no stolen flag inside a transaction output. A node checks whether a signature is valid, never whether the signer had any right to the key. The theft is cryptographically perfect and legally void at the same time. Everything outside the protocol works differently. Analytics firms can tag the exact outputs, exchanges can refuse the deposits, custodians can freeze accounts, and courts can order seizure. The coins stay spendable between two strangers and become close to unusable anywhere that checks identity. That gap is clearly why this may not be finished for the attacker. The Justice Department recovered 63.7 bitcoin from the Colonial Pipeline ransom, seized more than 94,000 from the Bitfinex theft, and took 50,676 from a Silk Road thief nearly a decade after the crime. None of it reversed a transaction. Each one required the holder to eventually touch something that asks who you are. For anyone still sitting on a seed made during that firmware window, the absence of a sweep is not evidence of safety. It may only mean your derivation path has not been scanned yet. Stay safe!!

• 
  HypeDojo (@HypeDojo) reported 44 minutes ago

  > Coinbase down. > Binance down. > Bitfinex down. Hyperliquid remained online. Market participants don't care about promises during bull markets.They care about performance when everyone shows up at once.

• 
  Ricardo Martinez (@ricardoXMR) reported 50 minutes ago

  @R3st4rtY0urL1f3 I don't think they are shutting down, but who knows. I know tether/bitfinex specifically chose to pull out of Europe to avoid MICA and the ****** EU & UK crypto regulations

• 
  Weaver (@Trend_Weaver) reported 54 minutes ago

  @bitfinex Been watching the same thing. Feels more like a slow grind than a strong reversal right now.

• 
  Bullish with CryptoKSA (@cryptowithKSA) reported 55 minutes ago

  ARGENTINA FREEZES 25 CRYPTO WALLETS IN $LIBRA PROBE 🚨 An Argentine judge ordered the freezing of 25 wallets linked to the $LIBRA investigation. Binance, Bybit, OKX, CoinEx, FixedFloat, and Bitfinex have also been ordered to provide KYC data and transaction records to help identify wallet holders and trace funds.

• 
  Gain (@GainMaxxing) reported 56 minutes ago

  @bitfinex Bitcoin is broken money, fiat is fake money.

• 
  SANDY.CRYPTOMAN (@SanMSH21) reported 56 minutes ago

  $Bitcoin ----Most Costly Mistakes The $44 Billion Bithumb Blunder (2026) A Bithumb employee accidentally sent 620,000 BTC instead of 2,000 Korean Won to users, forcing a massive, near-total network rollback and asset freeze. $44.000 Billion — Bithumb Error (2026) $9.000 Billion Mt. Gox Repayment Move (2024) $3.600 Billion DOJ Bitfinex Seizure (2022) $1.000 Billion DOJ Silk Road Move (2020) $0.003 B. Record Transaction Fee ($3.1M in 2023) $0.001 Billion Paxos Overpayment ($510,000 in 2023)

• 
  Erik (@ero_crypto) reported 58 minutes ago

  CENTRALIZED EXCHANGE ASSET FLOWS Where is capital moving across centralized exchanges this month? Looking at 30-day net flows across 65+ tracked CEXs, the picture shows a clear shift in capital and confidence between platforms. As of August 6, 2026, total CEX assets stand at $244.6B, down from $246.86B 30 days earlier — a net decline of $2.26B (-0.92%). ◾ 28 exchanges recorded inflows totaling +$893.5M ◾ 37 exchanges saw outflows totaling -$3.15B ◾ OKX: -$1.186B (-5.59%) ◾ Bybit: -$604.2M ◾ Bitfinex: +$263.91M ◾ Deribit: +$257.36M Bitfinex and Deribit attracted the strongest inflows relative to their size, while OKX and Bybit experienced some of the largest capital outflows. The broader takeaway: capital is leaving CEXs overall, but the flow between individual platforms is telling a much more interesting story.