What does your claim that AMMs would make money even if they were trading against arbitrage only have to do with the current crypto market?

Maybe my implication wasn't clear, but to put it another way, if the way you're making your bet is demonstrably terrible in an efficient market, it's probably not the best way to express your market view in an inefficient one either.

I'll admit I don't know what the implication here is supposed to be. It seems like you're implying that arbitrageurs aren't affected by high gas fees...

This is like the most super basic AMM stuff. Google "impermanent loss".

1. your post seemed to imply you cannot make money as an AMM LP if all flow is arbitrage. this is objectively incorrect. the fact that it may be suboptimal compared to some other much more difficult strategy, and/or one which has margin requirements, platform access requirements, or human time requirements, is not the gotchya you think it is.
2. yes that is correct.
3. This is like trying to figure out if a poker strategy is profitable but only counting the hands you lose. you have to net IL against fees. There is, in practice, a lot of bidirectional flow even in a trending market. This flow is even greater as there is more non-toxic flow, as "bad" trades create arbs for the toxic flow (this is easily seen by finding a high price impact DEX trade that goes through a single liquidity source). There is also a lot of CEX vs DEX arbitrage which is hard to identify as you can only see what happens onchain and can't tell if/when the DEX trader is out of the risk. Please show your work where IL is >> fees.

I'm curious what some think about how the Arbitrum narrative plays out & whether it's tangible FUD for Polygon over the summer (even if it's just perception and polygon's network effect continues to blossom). Are projects more easily portable now and we see a handful of DEXs, degen farms & ponzi/doges' on Arbitrum pretty quickly?

Will AAVE and the like be fast to deploy rewards programs paid out in wrapped Link over there? (I believe that's Arbitrum's gas token)

Other comments on sidechain/L2 state-of-affairs are much appreciated.

A naive high fee AMM will only make money (EV) against a small subset of arbitrage assuming there's informed flow elsewhere and the price isn't fundamentally stable or mean reverting like a stablecoin pair. You're basically saying the AMM makes money against arbitrageurs who are trading against fat fingers or unsophisticated whales dumping large market orders on the other side. In reality the AMM gets crushed by arbitrage when the market gaps or trends heavily.

My understanding is that's wrong, but I'm open to being corrected on that.


That's not the gotchya you think it is.

Me: You'll lose money in a trending market.
You: Yeah but if the drift is small relative to the chop you'll win.

No ****? You lose when the trend dominates. You win when it doesn't.


One thing you'll see when the markets mature and LP rewards inevitably either go to zero or a tiny fraction of what they are today, is a huge relative decrease in non-toxic flow. When ponzi tokens stop feeding the arbitrageurs and naive AMMs, the honeymoon ends, and the active market makers take over. Hopefully it'll eventually be possible to do that and maintain decentralization, but that seems pretty far off.

Sure they are. People trading shitcoins are not ordinary people who think they're sound investments, at least not for the most part. Though it's fair to say they don't understand the game they're playing or the extent to which they're gambling.

It isn't really much different, except that defi also has quite a few winners who got in early but aren't particularly sophisticated, at least for now.

can you elaborate on this one?

also i feel like the fees are prob a joke to the biggest players. just set ranges that make sense for the spreads and toss enough money where the fees are a low % compared to the yield generated in between range shifts. some dork that should have been hired by citadel but got scalped by some crypto fund is prob running the streets doing this, tho i must confess i'm not staring at liquidity charts 24/7

I'm curious about that too. I would assume "discriminate" means "dynamically adjust" as well, but "discriminate" certainly sounds like offering different spreads to different market participants.

I am upset to learn I will not get rich by just blindly throwing money at Uni V3 LPs

Casinos do pretty well. Even if defi is a shell-game (and it certainly is for a lot of what's happening on these defi protocols), I'm long casinos. Even if it's similar to Wall St. structured finance, there's tremendous demand for it for the foreseeable future, as it's a better game. I think some people want more risk (and so they enter, for various reasons), and some people want less risk, so they do all sorts of hedging and low-risk liquidity providing while these defi assets pop around in price.

We're all trying to figure out what money is because there's no stable asset - the dollar is inflating, gold is hard to hold or trust someone to do that for you, stocks and crypto pop around in price relative to each other, real estate is a headache when you have some real capital. If btc finds a stable value relative to the price of a house, car, dinner, or airplane ticket I'll be happy to exit the casino, but I don't expect everyone will, even in this outcome. For the short term though, smaller bankrolled individuals certainly can be empathized with for trying to "make it", by degenning into some sort of wealth -- as misguided as this might be. Work-hard-and-get-there is starting to appear untenable for a lot of peoples' situations.

I would add that some of the communities borne around these coins, perhaps especially the "meme coins", provide value in a way (if for a short time) that traditional ponzis do not. Maybe everyone gets rekt, maybe some meme themselves into a use-case and do what Gamestop seems to be doing.

Everything you've said here agrees with everything I said/linked before 🤝 And very much with my current portfolio and strategies.

Regarding mev-geth, the way it works is searchers submit transaction bundles in a private "fast lane" to miners representing 58% (last I checked) of the hashpower where there's a sealed price auction dynamic -- for the same arbitrage or liquidation or whatever MEV opportunity, the miner accepts the one with the highest bribe and includes it in the block as a 0 gwei transaction. Here is an example. Note, this was not the case even just a few months ago.


You can see the bribe that is transferred to the miner in the actual transaction.

This has had a couple interesting affects. First, it massively reduces network congestion as the txs that don't get included don't get broadcasted to the mempool (and then fail). The MEV searchers only compete with each other in their own tx pool. Second, it has made competing in the mempool far less profitable (and for many bots, unprofitable). Some mining pools doing their own extractions are not using Flashbots.

The normal geth client ordered the txs included in the block sorted by gas price (highest first), then for txs with the same gas price, it's random. This created a lot of spam from competition for profitable backrunning opportunities, and I believe it now sorts by time first seen: https://github.com/ethereum/go-ethereum/issues/21350.

Uniswap has been operating for many months now without any incentives on most of its trading pairs outside of the trading fees. The bull market has been kind to naive LPs and in many cases they're losing, but up in USD, and so they don't really feel it.

V3 is still so new and will take a bit of time for LP strategies to grow in complexity, but they're almost certainly negative EV a high % of the time without external incentives, and probably ~always in a bear market -- and will be even moreso for all but the very best (perhaps only MEV extractors, as I mentioned before) in 2-6 months.

Thanks for the explanation on that. I only had a vague idea of this MEV stuff previously.

As for this...

This MEV thing wouldn't mean that high gas fees are irrelevant to arbitrageurs. These MEV searchers still have to pay a "bribe" equivalent to at least the gas fee, so while their profit on any successful arb may not be affected by high gas fees, their overall profit is affected because smaller arbs aren't possible.

I think this is a bit overly cynical, but true for the majority of projects, for the simple fact that there are thousands of them at this point and the majority of those are low-effort copy pasta. However, I'm generally not doing a ton of altcoin trading unless I'm confident I've identified something specific that gives me an edge in the very near term (to sell immediately after). The majority of what I'm doing is related to: alpha related to knowledge of specific DeFi protocols, arbitrage, market making, carry trades, and collecting farm subsidies (generally to sell them for ETH and stablecoins, which I then compound into other strats, trying to keep my ~whole portfolio productive). At these levels, even the projects I like seem like a less good bet than ETH or USDC in a farm or making markets.

I think there's a really broad spectrum across all attributes here. Plenty of people know they're gambling, others think they're "investing", some are playing hot potato, others genuinely want to own certain assets. It really just depends on who you talk to, and to some extent, their view of where the market is at that moment in time (in an absolute sense, or alts/defi relative to eth/btc). Have heard serious people who genuinely think some top defi protocol tokens can still go 100x+ from here (which I think is really dumb)...

why do you think the fees are a joke? the main pools on Uniswap v3 have often done > 3x turnover per 24hrs. At 30bp fee, that's .9%/day gross (have to adjust for IL) for the *average* participant. It's a complex parimutuel game between all market participants. Just sort of picking a strategy randomly is giving money away. Ranges that "make sense" are easier to talk about and harder to actually generate because sometimes the market ranges for months and other times it full sends (2k to 4380 move was swift), then in 2 weeks 4380 -> 1750. Can be quite humbling and easy to get caught offsides or crushed by the trend (as pete mentioned).

yes, different spreads any "dumb retail" order flow can be given great execution. confident you will see this eventually

So literally discriminating based on the history of the address making the trade? Or are you suggesting some other metric?

The implications of this are interesting... ETH for example would lose some fungibility...

Yes so this is true -- if the arb opportunity is smaller than the tx fee from whichever tx(s) from the mempool would otherwise be included, then it won't get arbed -- as whichever txs are included, 0 gas price or not, do take up limited blockspace. The main difference though why it's so much less important than before Flashbots is MEV extraction txs that would otherwise fail (due to some other searcher winning the opportunity) never see the mempool or make it to the chain. In the recent past, there was a real cost to going for an opportunity and not winning it, as it results in a failed tx and wasted gas. This (plus growth of Polygon) is why we're seeing 25-50 gwei often these days when gas was very consistently 100-500 before Flashbots. Typically 500 gwei times are crazy volatile (lots of arbs/liquidations possible), but failing at 500 gwei is expensive!

When you compete in the mempool and get outbid, your tx fails, which not only costs you money and takes up blockspace, but the PGA you're taking place in drives up fees for everyone else trying to use the network, as bots bid up gas prices competing against each other, cause wallets to make bad fee estimations, etc. EIP-1559 changes this dynamic (for one, block size is elastic) and is much more efficient.

Parsec has an "MEV Extractions" element that is fun to look through for anyone interested. It is not free but if you spend any serious amount of time on this stuff, very worth it (the uniswap v3 visualization I posted earlier is also from parsec). I am just a very happy customer and not at all compensated for this.

that would be one such way; using a data graph

https://blog.synthetix.io/how-fee-re...-rebates-work/

You can read what synthetix does here (short: if you make a sharp trade you have to give your edge back). This is of course different because it's using an oracle (which is beatable bc latency), and because there's a delay in settlement which changes UX somewhat.

Regarding this, there were just two great podcast episodes on MEV. The first was on Bankless: http://podcast.banklesshq.com/66-cry...-charlie-noyes, with Phil Daian, who coined the term MEV in the first place and was the first person to even care about this stuff/wrote the initial flashboys 2.0 paper on DEX frontrunning, and Charlie Noyes and Georgios Konstanopolous who are two of the best minds in the space. The second episode is Uncommon Core with just Charlie and Georgios. Both worth a listen.

Never even thought of this, I'm more bullish on the future of AMMs now

Integral (a novel dex) is also using a delay

The flow is:
-user submits a trade with certain parameters
-after 5 minutes, your trade is executed at the 5 minute Uniswap v2 TWAP oracle price (so the time-weighted average price from time you submit trade until 5 minutes later, when it executes or reverts). If the price is outside the bounds you specify, it reverts.

This has some poor properties, a major one being that, aggregators can't route through it without making significant changes or taking on settlement risk themselves -- but could be a useful niche with lower fees and good for traders who just need to make a trade and want a fair price. Sustainably incentivizing LPs may be difficult (obviously a problem all AMM DEXs face in the longer term).

Any way you look at it, there is lots of experimentation and novel approaches being taken by DEXs and aggregators. Maybe it's all a giant scam and won't work though.

https://swap.archerdao.io/ is a noob-friendly way to avoid over-paying MEV - as far as I understand it (I'm quite confident I don't fully..)

You specify the transaction you want plus a miner tip (in place of a gas price, relates to how quickly you want your tx processed) and your order is routed through flashbots. So your order gets included in a block where an arbitrageur benefits from front-running you, you have no risk of massive slippage due to bots front-running your tx after seeing it in the mempool, or of your transaction failing due to slippage (which costs you gas).

Vitalik routed some of his shiba dump orders through archerdao, which brought it to my attention/seems like a fairly strong vouch. (Not long their gov token, fwiw)

The benefit of Archer is that you don't pay gas on transactions that would fail. You have no risk of massive price slippage if you simply specify in the tx parameters that you will not accept slippage-- and you can do this natively in all dex without Archer. Archer advertising this is "frontrunning protection" is very misleading. The frontrunning protection is in the DEX transaction parameters itself -- and nothing specific to archer's smart contract.

This is expanded on throughout this thread with hasu, Caleb (the archer founder), Robert Miller from Flashbots, and me.


Yet another great thing about Uniswap v3 is it eliminates the vast majority of profitable sandwich attacks by concentrating more liquidity near the mid-market price.

Wow that is amazing. Congrats to Caleb. Haven't talked to him in years

Thanks for the clarification.

Also found an article that expanded a little on the smart LP vs dumb LP strategies you were referencing above, I'm sure you're aware but for others following the conversation - https://research.parsec.finance/posts/amm-tradeoff

I heard that balancer were compensating LP providers with BAL tokens when they suffer IL from big discrete price jumps (which smart LP's could sidestep as described above) - are there any other AMMs which have tried to rectify this problem? Seems like a simple 1-4 block cooldown period for adding/removing LP tokens would render the strategy almost useless? I'm assuming the vast bulk of IL comes from wicks unwinding leveraged positions on CEXs, giving massive arb opportunities to bots.

^ is that from memento? good flick

I've got a huge question I would really appreciate if someone helped answer or pointed me to some resources. What is the deal with Cardano (and to a lesser extent other smart contract platforms) vs Ethereum? Attempting to google this leads to stuff like "Cardano is the Ethereum killer!" Is this really the goal of Cardano, or is this the equivalent of BTC moonbois?

I don't know enough about smart contract platforms to determine if there's room for more than one and Cardano/others are trying to carve out their own niche, or if July is supposed to be some kind of Highlander "There can be only one" scenario. If it's the latter, I can't envision first-mover advantage being overcome at this stage.

Cardano founder = Ethereum Co-Founder

He basically wanted to build a solid crypto from the ground up that improves everything about existing cryptos.

It's only starting to get going now so it has moon potential. (100x+)

I don't believe in a winner takes it all market.

BTC = crypto gold, ETH = currently most widely used blockchain, Cardano might be the next improved ETH. I believe they can co-exist, but time will tell...