Uniswap V3 price analysis

In this notebook we will show how to download price events from Uniswap V3 to your computer as CSV files and use them to analyse price in each pool.

For more background information, see this blog post about the topic
You need to understand Jupyter Notebook and Pandas basics
You need to understand Ethereum and Web3.py basics
You will need to have Ethereum API node and its JSON-RPC URL in order to pull out the data from Ethereum blockchain. The notebook will interactively ask you for your API key.
You can click Launch binder button on this documentation page to automatically open this notebook to be run in Binder cloud notebook service. Note that we recommend to run the notebook on your local computer, as generated CSV needs are large and take long time to generate.

Download the raw data from Ethereum blockchain

For simplicity, you can sign up for free access to an Ethereum node for example at Infura, however we recommend you to run your own Ethereum node.

[10]:

from web3 import Web3, HTTPProvider

# Get your node JSON-RPC URL
json_rpc_url = input("Please enter your Ethereum mainnet JSON-RPC URL here")
web3 = Web3(HTTPProvider(json_rpc_url))

As an example, here we download raw events from first few blocks after Uniswap V3 was deployed. The events will be stored in several CSV files at /tmp folder.

Depends on your internet connection and latency to the Ethereum node, the scan might take hours. However it can resume in case there is a crash, as we save the last scanned block in a JSON state file.

[11]:

from eth_defi.uniswap_v3.constants import UNISWAP_V3_FACTORY_CREATED_AT_BLOCK
from eth_defi.uniswap_v3.events import fetch_events_to_csv
from eth_defi.event_reader.json_state import JSONFileScanState

start_block = UNISWAP_V3_FACTORY_CREATED_AT_BLOCK
end_block = UNISWAP_V3_FACTORY_CREATED_AT_BLOCK + 250_000

# Stores the last block number of event data we store
state = JSONFileScanState("/tmp/uniswap-v3-price-scan.json")

fetch_events_to_csv(json_rpc_url, state, start_block=start_block, end_block=end_block)

Restored previous scan state, data until block 12,619,621, we are skipping 250,000 blocks out of 250,000 total
Scanning block range 12,619,621 - 12,619,621

Wrote 0 PoolCreated events
Wrote 0 Swap events
Wrote 0 Mint events
Wrote 0 Burn events

Analysing Uniswap v3 price formation

In Uniswap V3, you can get the current price of any pool from any given moment using swap events.

[12]:

import pandas as pd

swap_df = pd.read_csv("/tmp/uniswap-v3-swap.csv")

print(f"We have total {len(swap_df):,} swaps in the dataset")

We have total 1,229,361 swaps in the dataset

Choose a pool to analyse, for example: USDC/ETH 0.3%. You can find pool addresses on Uniswap v3 info site.

[13]:

from eth_defi.uniswap_v3.pool import fetch_pool_details

pool_address = "0x8ad599c3A0ff1De082011EFDDc58f1908eb6e6D8"
pool_details = fetch_pool_details(web3, pool_address)

print(pool_details)
print("token0 is", pool_details.token0, "with", pool_details.token0.decimals, "decimals")
print("token1 is", pool_details.token1, "with", pool_details.token1.decimals, "decimals")

Pool 0x8ad599c3A0ff1De082011EFDDc58f1908eb6e6D8 is USDC-WETH, with the fee 0.3000%
token0 is <USD Coin (USDC) at 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48> with 6 decimals
token1 is <Wrapped Ether (WETH) at 0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2> with 18 decimals

Extract the swaps of this pool from the Uniswap v3 full dataset of all pools.

[14]:

df = swap_df.loc[swap_df.pool_contract_address == pool_address.lower()]
df

[14]:

	block_number	timestamp	tx_hash	log_index	pool_contract_address	amount0	amount1	sqrt_price_x96	liquidity	tick
19	12371376	2021-05-05T01:56:23	0xce7c3c307d820785caa12938012372fc9366a614a6aa...	26	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	-329608	100000000000000	1377932816571815120446551350158799	4303369674465501	195285
21	12373132	2021-05-05T08:23:26	0x9a1c51b0bffbf840948f3b6e3f3e495ba1cd3fa64854...	192	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	-164694492	50000000000000000	1378850591292581266780357299649652	4303369674465501	195298
25	12373520	2021-05-05T09:50:51	0xc58715c62a5bf70a6ca09f0e51546d6cad76c8d4fff0...	8	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	-329169	100000000000000	1378852426842022799073024911548633	4303369674465501	195298
26	12374077	2021-05-05T11:59:57	0x288c21b8b4fbf449b1d086a06e43b124ac2bc088c3f5...	86	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	2	-329169	1378852426842016741051966412054516	4304946248093346	195298
27	12374320	2021-05-05T12:56:56	0x67502d8ba373287f6d301f6baa77c5a5f4c80d0753c3...	257	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	1559137299	-467880854065813753	1370241555019945317645788135487819	4304946248093346	195173
...	...	...	...	...	...	...	...	...	...	...
1229165	12619596	2021-06-12T12:16:13	0x8c1f6c5639b304eb0087c2b6d5b65e32ee4f1a51e8b0...	112	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	92432639474	-38329545577539549888	1615736234513681125600943820478722	25998097729965548918	198469
1229184	12619598	2021-06-12T12:16:25	0xac8e34aa704a1ab4068bd93d4b5aef0fb7655d292fc9...	3	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	76760509615	-31826486546079303210	1615639241428978886102798535892145	25997255949450154096	198468
1229186	12619598	2021-06-12T12:16:25	0x0752c6529f939f046c7f9648e8d27d76b1cf767d6902...	8	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	482579746068	-200000000000000000000	1615029729696907486273078115160620	25997255949450154096	198460
1229190	12619598	2021-06-12T12:16:25	0xbd9193ff70e4828a43679f955c9707fa63fe1dabeed3...	45	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	679981136491	-281554906013205614934	1614171674604720940565192048438458	25997255949450154096	198449
1229350	12619619	2021-06-12T12:21:51	0x91f7b948333f452324270b09d6cb074b69dda7647f4c...	281	0x8ad599c3a0ff1de082011efddc58f1908eb6e6d8	8388626	-3471572647773651	1614171664025625771996673894396803	25999040327671324243	198449

64090 rows × 10 columns

In Uniswap V3, we can get pool’s price from pool’s tick.

[15]:

def convert_price(row):
    # USDC/WETH pool has reverse token order, so let's flip it WETH/USDC
    tick = row["tick"]
    return pool_details.convert_price_to_human(tick, reverse_token_order=True)

def convert_value(row):
    # USDC is token0 and amount0
    price = float(row["price"])
    return abs(float(row["amount0"])) / (10**pool_details.token0.decimals)

df = df.copy(deep=True)  # https://stackoverflow.com/a/60885847/315168
df["price"] = df.apply(convert_price, axis=1)
df["value"] = df.apply(convert_value, axis=1)
df[["block_number", "timestamp", "tick", "price", "value"]]

[15]:

	block_number	timestamp	tick	price	value
19	12371376	2021-05-05T01:56:23	195285	3306.010571	0.329608
21	12373132	2021-05-05T08:23:26	195298	3301.715764	164.694492
25	12373520	2021-05-05T09:50:51	195298	3301.715764	0.329169
26	12374077	2021-05-05T11:59:57	195298	3301.715764	0.000002
27	12374320	2021-05-05T12:56:56	195173	3343.244146	1559.137299
...	...	...	...	...	...
1229165	12619596	2021-06-12T12:16:13	198469	2404.538795	92432.639474
1229184	12619598	2021-06-12T12:16:25	198468	2404.779249	76760.509615
1229186	12619598	2021-06-12T12:16:25	198460	2406.703745	482579.746068
1229190	12619598	2021-06-12T12:16:25	198449	2409.352444	679981.136491
1229350	12619619	2021-06-12T12:21:51	198449	2409.352444	8.388626

64090 rows × 5 columns

Then we can convert linear price data to OHLC candles.

[16]:

from eth_defi.research.candle import convert_to_ohlcv_candles

candles = convert_to_ohlcv_candles(df, time_bucket=pd.Timedelta("4h"))

# Show only 100 first candles
candles = candles.head(100)

candles

[16]:

	open	high	low	close	volume
timestamp
2021-05-05 00:00:00	3306.010571	3306.010571	3306.010571	3306.010571	3.296080e-01
2021-05-05 04:00:00	NaN	NaN	NaN	NaN	0.000000e+00
2021-05-05 08:00:00	3301.715764	3301.715764	3301.715764	3301.715764	1.650237e+02
2021-05-05 12:00:00	3343.244146	3343.244146	3343.244146	3343.244146	1.559137e+03
2021-05-05 16:00:00	3343.244146	3476.895080	3343.244146	3430.617750	1.393686e+05
...	...	...	...	...	...
2021-05-20 20:00:00	2789.467427	2890.845172	2668.863966	2775.833150	1.082835e+08
2021-05-21 00:00:00	2775.833150	2924.572530	2775.833150	2797.008788	4.045975e+07
2021-05-21 04:00:00	2792.816630	2809.342124	2694.068302	2694.068302	4.383654e+07
2021-05-21 08:00:00	2691.375715	2771.395590	2638.085543	2707.571699	4.017013e+07
2021-05-21 12:00:00	2709.738515	2750.138850	2388.005409	2420.944555	1.300869e+08

100 rows × 5 columns

Drawing OHLCV chart

Now we can plot the OHLC chart using Plotly.

[17]:

import plotly.graph_objects as go
from plotly.subplots import make_subplots

candlesticks = go.Candlestick(
    x=candles.index,
    open=candles['open'],
    high=candles['high'],
    low=candles['low'],
    close=candles['close'],
    showlegend=False
)

volume_bars = go.Bar(
    x=candles.index,
    y=candles['volume'],
    showlegend=False,
    marker={
        "color": "rgba(128,128,128,0.5)",
    }
)

fig = go.Figure(candlesticks)
fig = make_subplots(specs=[[{"secondary_y": True}]])
fig.add_trace(candlesticks, secondary_y=True)
fig.add_trace(volume_bars, secondary_y=False)
fig.update_layout(title="ETH/USDC pool after Uniswap v3 deployment", height=800)
fig.update_yaxes(title="Price $", secondary_y=True, showgrid=True)
fig.update_yaxes(title="Volume $", secondary_y=False, showgrid=False)
fig.show()