So in other words I was exactly right in what I said, with one qualifier:

Unless you're splashing around in shitcoins on BSc -- which obv comes with its own rug pull risks -- LP is a whale-only game because a) gas fees destroy any deposit sub 1 ETH, and b) LP rewards are distributed based on how much liquidity you provide to the pool.

It's far easier, wiser, and more profitable long-term for all but a few hardcore traders with at least 100K bankrolls (and I think that's being really generous tbh) to simply stake.

Staking requires gas to enter and leave and provides a yield
Liquidity providing requires gas to enter and leave and provides a yield

I guess I don’t see the difference from a expense point of view. Staking is simpler though and you’re correct that for most people it is a far easier and more sensible option - but that’s nothing to do with being a whale or not, just that most people won’t understand the maths of liquidity providing.

Honestly, with even $1m roll ethereum will be quite expensive for you. Your best bet will be to choose your pools wisely such that you can stay in them for reasonable amounts of time. Jumping around pools daily with $100k is suicide and with $1m is quite tough. Also, gas fees tend to be a lot lower on the weekends, so try to make changes/claims etc as needed on the weekend.

0xdef1 created this calculator for optimal compounding which you should find helpful:

You can also adapt it by altering the formulas to make it weekly, monthly, etc instead of yearly.

The good news is, as of this morning the block gas limit was raised to 15M from 12.5M which is a small help, with some bigger help coming soon with Optimism (fingers crossed no more delay).

I don’t jump around daily or even weekly since I have a day job but I’m making excellent yields on liquidity providing, and I think it is perfectly possible with 100k to do the same.

I've been fooling around on BSC for a while farming and have finally got onto Ethereum. Where is the best returns right now on just ethereum? Curve seems decent, but I am sure there are better ones out there.

Alpha leak: In general LPing 2 assets you're [correctly] bullish on is quite good. https://research.paradigm.xyz/uniswaps-alchemy

The only real assumption here is that you have ~max[log(wealth)] preferences which is true for most people. The main type of IL you should worry about is when you're LPing shitcoins that are very volatile, especially to the downside. Even more true for scam projects that get rugged by the team. In the other case where one asset goes up a lot more than the other, it's just automated profit taking with a fee on each trade -- not the worst result in the world. Especially true if it's a relatively illiquid asset you own a lot of, whereby selling it into the pool = massive slippage.

Of course, you do need to be mindful about how much gas you're spending, including the necessary approval txs, as well as the eventual remove, possible sells on the backend after remove, and slippage from the backend sale. If you don't know what I mean by that, don't LP, and learn first.

As a rule of thumb (for the sake of these examples you can assume the 2nd asset is stable, or otherwise that the delta is in the crossprice if LPing 2 volatile assets, say, COMP/ETH):
One asset doubles -> 5.7% IL
One asset 5x's -> 25.5% IL
One asset halves -> -5.7% IL
One asset goes to 0 -> 100% loss

Note for IL, it compares LPing vs holding the same basket of assets without LPing. 25% IL (one asset flat, one 5x's) doesn't mean you lost money in an absolute sense -- just that you would have done better if you didn't LP, because you synthetically sold the better performing asset all the way up.

Now, what this *does not* factor in is the 60bp spread Uniswap LPs charge (variable on Balancer). In order for AMM price to change, there has to be trades (this is literally how Uniswap works; the price is deterministic and only changes as a result of interactions with the pool). Trades -> fees to LPs. This is the key. When a price moves (let's say, doubles), it doesn't typically teleport from 1 to 2 in a straight line. Instead it chops there. For simplicity, let's say it goes to 1.10, then 1.05, then 1.15, etc. If you view this on a chart it looks like this:



Now if you look between the red lines there is a trade down (sell) and [fraction of] a trade up (buy). On both of these, the LP charges 30bp fee. So if the trade down was $10,000 and the trade up was $20,000, the LP makes ~.3%*$30,000 = $90. These fees help compensate for IL. This is "volatility harvesting".

If you take the `[24h trade volume]/[pool liquidity]*.3*365` that will give you the annual yield (assuming of course that volume and liquidity stay constant, which they will not). There is a lot of edge in estimating these things properly.

As a rule of thumb, a pool that does:
.5x turnover = 54.75%
1x turnover = 109.5%
... and so on. And any yield you receive from staking on top of that would be added.

Also a beautiful thing about Uniswap v2* LP tokens is they auto-compound every trade without any intervention needed from the LP. This is not true for the subsidy portion (which is typically yield in a token in exchange for staking your LP tokens -- in order for that to compound, you'd have to periodically claim + add it to the pool, incurring more and more gas fees each time you interact with a smart contract).

*this is different in v3, as LP tokens are non-fungible.

Cliffs: do the math. don't decide if it's profitable or not based on your feelings and a few days of data.

It definitely is possible -- just saying you can't run the same strategy as someone with $10m or $100m.

In general I'd say LPing any position less than $20k on ethereum layer 1 is almost definitely a waste now. $20-100k often fine but can still be iffy. Price the gas fees into your trades. If approve -> pool -> stake -> unstake -> approve remove -> remove is more than x% of your position, avoid. Up to you to figure out what x is; it obviously depends on the trade volume, subsidy, how long it lasts etc.

Also be mindful of how taxes affect you. The commonly accepted position is that providing uniswap liquidity = taxable event, in which you sell the tokens you are providing in exchange for the LP tokens. Removing them would also be a taxable event, where you sell the LP tokens back for 50% token A, 50% token B.

(I am not a tax professional but this is what US-based funds etc are doing)

I think you're underestimating a) how advanced you are (@Iknownothing) -- this is a steep learning curve -- and b) how much retail traders are priced out of ETH DEX nowadays as far as proper bankroll management is concerned wrt LPing/gas fees/etc.

Thanks Two SHAE, very informative.

This was my problem - I was attracted by the high yields but didn't understand IP well enough.

Personally, I don't have the time (or inclination) to do the maths, in-depth research, or actively manage any positions. So I think my best bet is just to stick to staking - set and forget.

Im planning on running a RocketPool node - is anyone else here running ETH validators?

Best to just use matic with a smaller roll?

Great place to start IMO. Potentially a decent place to hang out this year.

I'm pretty bullish on matic in the mid-term. A lot of people more knowledgeable than me have been saying to wait for other protocols to widespread launch and crush everything - SOL, ETH 2.0/rollups/etc, DOT, but what I've seen so far is just literally the easy-onboard uses get traction. We are in a bull market and retail doesn't care about decentralization, they just want to feel like they are "doing crypto" (hence binance smart chain's success). Gas fees on matic/polygon are next to nothing, you can port over from ETH mainnet as well as bsc, transactions are lightning fast even in market flash-crashes (so far), and plenty of projects are porting form eth mainnet on over here.

hot take: Impermanent Loss most overrated thing in crypto

If you have a small BR, buy and HODL and spend 6-12 months getting up to speed. Earn those stripes.

the real problem is when you LP a shitcoin with ETH or USD and your farm is in that shitcoin.
Once your shitcoin goes down you lose money in the liquidity pool (and you automatically buy more of it!) and your farm becomes worthless.

Yup. It's like an exotic derivative. Your LP position is long vol, very long shitcoin delta, but short gamma (but this is partially offset by the shitcoin yield). The far downside case is the grossest where you lose 2x your size in the shitcoin instead of 1x.

An interesting use case which is necessary for well-functioning markets is the ability to short the shitcoins via DeFi (which requires a lending facility). This also allows delta/gamma hedging for LPs. Ever wonder why FTX lists futures so fast?

I think you need to be doing $100k+ on ETH to be worth it. Would just do terra or BSC if LPing smaller than that.

I was worried I was too dumb for liquidity pools. Turns out I’m also too poor.

.

Generally agree. To give a little more color, here's some quick and dirty guidelines for what opportunities are practically available to you on ethereum given a particular position size. The bigger your roll, the more transaction costs become a rounding error and the more you are able to be nimble in managing your positions.

100k+
- LP/farm for days
ie. pool 2 for a new token like FORTH

30k+
- LP for months
ie. WBTC/ETH or a Curve pool

10k+
- lend or stake for months
ie. lend USDC on Alpha or vault BTC on Vesper or Keeper

I recommend going old school and opening up a spreadsheet and running the math for whatever you want to do. As TwoSHAE mentioned, include all costs for approve, supply, remove, swap on backend, etc. As an LP you are selling volatility, so I'd at least learn the super basics of options pricing and the greeks.

It's interesting that Aave decided to build out on Matic. They obviously had the option to build on Optimism along with SNX, UNI, COMP, etc. It's unclear why they would forego the composability offered there. That's either a big vote of confidence for Matic or some doubts about Optimism.

I've been having doubts about Optimism myself. The delay isn't great and it's unclear how smoothly the 7-day lockup will be dealt with. It has some potential to be a letdown.

The more I listen to devs smarter than me, the more it sounds like ZK rollups are a very attractive scaling solution long term. They are just much faster, more efficient, and more powerful. Of course the math is hard as hell and it is a mess to migrate there. IMO dydx is killing it with their StarkWare solution and this is a big proof of concept for zk rollups.

I wish I was smart enough to understand the differences between all those sidechains.

xDai and matic look identical to me for example (except for the different farms on it ofc)

Doesn't have all that much to do with intelligence. Much more to do with exposure & experience; discipline and hard work; and research.

Vitalik talks about this all the time. YouTube. Google his blog post. Increasingly devote yourself to the space for long enough and you learn.

It’s hard to ignore that feeling of missing out, missing the next big thing what a disaster...

But I’m not a technologist or computer science major and I shouldn’t try to be. Win the battle you’re equipped for and savour that victory - and cheer other people winning their own victories.

Ps Your monthly tips for yield farming are what got me started down that road - thank you very much!

thank you, I think I only did one or two? Anyways, one month is way too long, the game changes every week on ETH. And the best farm on BSC can change daily.

The things I learned recently are:
A) It is pretty incredible how hard APYs can drop when the market goes down. The coins you farm will drop much harder than BTC or ETH.
B) Curve is pretty much unusable if you don't have a 7 digit bankroll when fees are high (100+ gwei), but when fees are non-existent (polygon) or low (50-60 gwei on ETH right now) it is amazing. APYs are even higher than shown on the website if you harvest and reinvest somewhat frequently.

edit: I can make a farming guide for May, but don't expect it to be helpful for long.