Competitor AMMs KPIs #61

PoloX2021 · 2024-11-06T15:05:28Z

This PR measures TVL, 24H Volume and 24H APR for 4 existing AMMs (Uniswap, Sushiswap, Pancakeswap and curve)
The 24H APR is defined as sum(volume*feePct/TVL)*365 over the past 24H.

Uni, Sushi and Pancake are based on the same code. The queries are very similar to one another with only differences in the tables to look into.
For each, there is one first query which looks at the reserves of each pool, to find the largest ones.
Then the query sync comes from previous code and kpis is a modified version of the 10k comparison to measure only the 24H APR.

For Curve, there are less tables available and information in the events. So, we calculate the reserves based on all transfers of the 2 tokens involved in the pools.
The trades are dervived from a dedicated dune table.
There might be one issue with ETH transfer which seems to impact only one pool with appears to have a negative TVL.

…sfers

fleupold

Can you please run sqlfluff fix to make sure the sql is formatted nicely (this makes it easier for me to review). Also, I'd personally prefer smaller PRs (e.g. one per project) as then I can review them 1 by 1 in 20 minutes each (rather than having to block 1h for the whole PR, which at least for me happens less frequently).

As for the structure, the /cowamm currently contains two folders (kip and profitability). Inside profitibality, we now add competitor_kpis. I find this confusing.

Can we either create /cowamm/competitors/kpi or /comamm/kpi/competitors?

More comments (mainly about semantic inconsistencies inline).

cowamm/profitability/competitor_kpis/curve/curve_kpis_4232873.sql

cowamm/profitability/competitor_kpis/curve/curve_largest_2token_pools_4232976.sql

fleupold · 2024-11-09T03:28:56Z

cowamm/profitability/competitor_kpis/curve/curve_largest_2token_pools_4232976.sql

+coin1 as token1,
+mid_fee*power(10,-10) as fee
+from curvefi_{{blockchain}}.view_pools
+where coin2 = 0x0000000000000000000000000000000000000000),


Why is this needed? Intuitively this looks to me as if this is an invalid pool (with coin2 being the 0 address). Could you add a comment in code please?

Curve can have multi-tokens pools, which would not make sense to compare with CoW AMM for now

cowamm/profitability/competitor_kpis/pancakeswap/pancakeswap_largest_pools_4232597.sql

cowamm/profitability/competitor_kpis/pancakeswap/pancakeswap_kpis_4232738.sql

cowamm/profitability/competitor_kpis/pancakeswap/pancakeswap_syncs_4232660.sql

cowamm/profitability/competitor_kpis/sushiswap/sushiswap_kpis_4232588.sql

…s comments

…the same query

fleupold

Looking very good. Would like to see a way to distinguish between different uni v2 like protocols though as this is very likely something the UI wants to display.

cowamm/kpi/competitors/curve/curve_largest_2token_pools_4232976.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_kpis_4304295.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_syncs_4304212.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_kpis_4304295.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_syncs_4304212.sql

shoom3301 · 2024-11-22T10:11:45Z

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_kpis_4304295.sql

+
+select
+    contract_address,
+    0.003 as fee,


Fee is hardcoded to 0.003 here, but some pools might have different fee values, don't they?

Thanks for the comment! Pancake Swap is only .25%
For Sushi and Uni swap it is harcoded to .3%.

fleupold

Very close. I just don't understand the self join in the curve query. The rest are nits.

fleupold · 2024-11-25T12:33:20Z

cowamm/kpi/competitors/curve/curve_kpis_4232873.sql

+from "query_4232976(blockchain='{{blockchain}}')" as r
+inner join "query_4232976(blockchain='{{blockchain}}')" as r1


Why is this self join needed? Can you please document it in the code.

fleupold · 2024-11-25T12:33:40Z

cowamm/kpi/competitors/curve/curve_kpis_4232873.sql

+        and r.tx_hash = t.tx_hash
+where
+    t.block_time >= date_add('day', -1, (case when '{{end_time}}' = '2024-01-01' then now() else timestamp '{{end_time}}' end))
+    and r1.latest = 1


4232976 already filters for latest = 1 doesn't it?

No, it might feel confusing in the last part of the code: it filters by latest_per_pool only to find the largest pools, then it includes all of the tvls after every event which happened

select * from recent_tvl where contract_address in ( select contract_address from recent_tvl where latest_per_pool = 1 order by tvl desc limit {{number_of_pools}} )

cowamm/kpi/competitors/curve/curve_largest_2token_pools_4232976.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_pools_4303563.sql

cowamm/kpi/competitors/uni_swap_style/largest_uni_style_kpis_4304295.sql

fleupold

👏 Looks great, only nits remaining

cowamm/kpi/competitors/all_competitors_4335231.sql

fleupold · 2024-11-26T15:46:17Z

cowamm/kpi/competitors/curve/curve_kpis_4232873.sql

+    sum(amount_usd) as volume,
+    365 * sum(amount_usd * r.fee / r.tvl) as apr


Thanks for the explanation, I get it now. Just as a comment (not actionable assuming performance of the join isn't an issue), but this could also be computed using a "rolling sum" which would avoid the self join.

Something like

select contract_address, fee, tvl, sum(amount_usd) over (partition by contract_address order by latest_per_pool) as volume, 365 * sum(amount_usd * fee / tvl) over (partition by contract_address order by latest_per_pool) as apr

There is indeed a performance issue overall for all queries. So I implemented it. Result: it divides by 2 the execution time

fleupold · 2024-11-26T15:48:29Z

cowamm/kpi/competitors/curve/curve_largest_2token_pools_4232976.sql

+        sum(transfer0) over (partition by contract_address order by block_time, evt_index) as reserve0,
+        sum(transfer1) over (partition by contract_address order by block_time, evt_index) as reserve1,


Looking at the performance of the query and the fact that we are already joining with prices for sell and buy token here, I wonder if we wouldn't dramatically speed up this query by also computing the amount_usd per tx_hash here instead of computing it in the other query.

Not sure if it would speed up or not, because we would have to look for the transaction events and not simply the transfers

fleupold · 2024-11-26T15:51:29Z

cowamm/kpi/competitors/all_competitors_4335231.sql

+
+--Curve
+select
+    contract_address,


surprised you don't need the token addresses as well here (would make sense imo)

Sasha was happy with all the columns in the table...

Co-authored-by: Felix Leupold <[email protected]>

PoloX2021 added 3 commits November 6, 2024 14:48

New queries to measure the kpis from competitors

fcdb0bd

new parameter added because of the previous queries

7bb4668

Excludes possible negative tvls because of issues with ERC20/ETH tran…

dd0dccc

…sfers

PoloX2021 assigned fleupold, fhenneke, harisang and bram-vdberg and unassigned fleupold, fhenneke, harisang and bram-vdberg Nov 6, 2024

PoloX2021 requested review from fleupold, fhenneke, harisang and bram-vdberg November 6, 2024 15:30

PoloX2021 added 2 commits November 6, 2024 16:10

Formatting

45d3f2a

Added a missing blockchain parameter

752bf52

harisang changed the title ~~Comptetitor AMMs KPIs~~ Competitor AMMs KPIs Nov 8, 2024

fleupold reviewed Nov 9, 2024

View reviewed changes

PoloX2021 added 8 commits November 15, 2024 11:56

Simplified the querying of all tranactions from a curve pool

9fc431a

Few comments and adapting to remarks for Curve

7060d6d

Added filter by x largest pools

db8459a

Added/propagated through queries number_of_pools parameter and variou…

cb41f52

…s comments

Removed the union for inactive pools. THe left join does the work

27ac2db

Fixed the sqlfluff

1f328eb

Moved the folder to kpis/competitors

0bf6622

New query to replace the uni style querying. they are now grouped in …

0d0f7f5

…the same query

fleupold reviewed Nov 21, 2024

View reviewed changes

shoom3301 reviewed Nov 22, 2024

View reviewed changes

Added the project value to classify the pools: uni/ pancake/sushi

72f5c6a

PoloX2021 added 5 commits November 22, 2024 13:32

Comments

73a55b2

Added parameter for time window, and removed avg(tvl), avg(fee)

77c8121

upadated default parameter

a6478fa

Added syncs inline to kpis.

d53d5c0

Added flag uni/pancake/Sushi and parameter for time window

7f4dbb9

fleupold reviewed Nov 25, 2024

View reviewed changes

PoloX2021 added 3 commits November 26, 2024 10:10

New default value for day, few comments and renaming

5b26df3

Changed deployer contract addres for uniswap

cd66d9f

Query to regroup all the different projects kpis together

694d5cf

fleupold approved these changes Nov 26, 2024

View reviewed changes

PoloX2021 and others added 3 commits November 26, 2024 16:09

Typo for chain_id

df49fb1

Co-authored-by: Felix Leupold <[email protected]>

fixed typo and added base

f162872

updated kpis to filter before joining and improve performance

f08a4b5

PoloX2021 merged commit 7b60835 into main Nov 27, 2024
1 check passed

PoloX2021 deleted the comptetitor-APRs branch November 27, 2024 13:04

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Competitor AMMs KPIs #61

Competitor AMMs KPIs #61

PoloX2021 commented Nov 6, 2024

fleupold left a comment

fleupold Nov 9, 2024

PoloX2021 Nov 15, 2024

fleupold left a comment

shoom3301 Nov 22, 2024

PoloX2021 Nov 22, 2024

fleupold left a comment

fleupold Nov 25, 2024

fleupold Nov 25, 2024

PoloX2021 Nov 26, 2024

fleupold left a comment

fleupold Nov 26, 2024

PoloX2021 Nov 26, 2024

fleupold Nov 26, 2024

PoloX2021 Nov 26, 2024

fleupold Nov 26, 2024

PoloX2021 Nov 26, 2024

		from "query_4232976(blockchain='{{blockchain}}')" as r
		inner join "query_4232976(blockchain='{{blockchain}}')" as r1

		sum(amount_usd) as volume,
		365 * sum(amount_usd * r.fee / r.tvl) as apr

		sum(transfer0) over (partition by contract_address order by block_time, evt_index) as reserve0,
		sum(transfer1) over (partition by contract_address order by block_time, evt_index) as reserve1,

Competitor AMMs KPIs #61

Competitor AMMs KPIs #61

Conversation

PoloX2021 commented Nov 6, 2024

fleupold left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fleupold left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fleupold left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

fleupold left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment