initial

calculate_profit_from_logs.py

@@ -0,0 +1,107 @@
+import pygraphviz as pgv
+import csv, csv_hack, os
+import json
+from exchanges import get_trade_data_from_log_item
+
+COLORS = ["red", "blue", "green", "orange", "purple", "black", "yellow", "grey", "darkgreen"] * 10
+
+logsdict = csv.DictReader(open('data/all_logs_bigquery.csv'), delimiter=',',
+                            quotechar='"', quoting=csv.QUOTE_MINIMAL)
+logs = {}
+
+def get_rate_label(token1, amount1, token2, amount2):
+    if (token1 >= token2 and token1 != "ETH" and token1 != "WETH") or (token2 == "ETH" or token2 == "WETH"): # arbitrary ordering/ tiebreak
+        try:
+            return "%4g %s" % (amount1/amount2, token1 + '/' + token2)
+        except ZeroDivisionError:
+            return "[INF] %s" % (token1 + '/' + token2)
+    try:
+        return "%4g %s" % (amount2/amount1, token2 + '/' + token1)
+    except ZeroDivisionError:
+        return "[INF] %s" % (token2 + '/' + token1)
+
+
+def get_profit_graph(logset, txhash):
+    dot = pgv.AGraph(label=txhash + ' Profit Flow', directed=True, strict=False, nodesep=1.0, ranksep=0.5, sep=0.0, labelfloat=False)
+    unknown = False
+    graph_edges = []
+    logindex = 1
+    tokens_involved = set()
+    trades = []
+    for logitem in logset:
+        address = logitem[0]
+        data = logitem[1]
+        topicstext = logitem[2].replace('\'', '\"')
+        topics = json.loads(topicstext)
+        data = data[2:] # strip 0x from hex
+        trades_data = get_trade_data_from_log_item(topics, data, address)
+        if trades_data is not None:
+            for trade_data in trades_data:
+                (tokenget_addr, tokenget_label, tokenget, amountget, tokengive_addr, tokengive_label, tokengive, amountgive, exchange) = trade_data
+                graph_edges.append((tokenget, "!" + exchange, amountget)) # (add "!" to mark special exchange node)
+                graph_edges.append(("!" + exchange, tokengive, amountgive))
+
+                rate_label = get_rate_label(tokenget, amountget, tokengive, amountgive)
+                tradenode_label = "Trade #" + str(logindex) + " (" + exchange + ")\n" + rate_label
+                dot.add_edge(tokenget_label, tradenode_label, label=("%4g" % amountget), color=COLORS[logindex])
+                dot.add_edge(tradenode_label, tokengive_label, label=("%4g" % amountgive), color=COLORS[logindex])
+                trades.append(tradenode_label)
+                tokens_involved.add(tokenget_label)
+                tokens_involved.add(tokengive_label)
+                logindex += 1
+        else:
+            # some item in the logset failed to parse => we don't have complete profit picture
+            unknown = True
+    for token in list(tokens_involved):
+        dot.add_subgraph(token, rank='same')
+    dot.add_subgraph(trades, rank='same')
+    for i in range(0, len(trades) - 1):
+        dot.add_edge(trades[i], trades[i+1], style="invis")
+    return(graph_edges, unknown, dot)
+
+def calculate_profit_for(profit_graph):
+    token_profits = {}
+    for edge in profit_graph:
+        if not edge[0] in token_profits:
+            token_profits[edge[0]] = 0
+        if not edge[1] in token_profits:
+            token_profits[edge[1]] = 0
+        token_profits[edge[0]] -= edge[2]
+        token_profits[edge[1]] += edge[2]
+    return token_profits
+
+for log in logsdict:
+    hash = log['transaction_hash']
+    if not hash in logs:
+        logs[hash] = []
+    logs[hash].append((log['address'], log['data'], log['topics']))
+
+
+spamwriter = csv.writer(open('data/profits.csv', 'w'), delimiter=',',
+                            quotechar='"', quoting=csv.QUOTE_MINIMAL)
+spamwriter.writerow(["txhash","drawn","unknown","all_positive","eth_profit","profit_graph","profit_calcs"])
+
+i = 0
+total = len(logs)
+for txhash in logs:
+    i += 1
+    print(txhash, i, "of", total)
+    output_file_name = 'profit_graphs/' + txhash + '.png'
+    drawn = False
+    (profit_graph, unknown, dot) = get_profit_graph(logs[txhash], txhash)
+    if unknown:
+        # failed to process given entry because some exchange that's in dex_list.py is missing a log parser
+        print("UNKNOWN!", txhash)
+    if not unknown and len(profit_graph) > 2:
+        if not os.path.exists(output_file_name):
+            dot.draw(output_file_name, prog="dot")
+        drawn = True
+    profit_calcs = calculate_profit_for(profit_graph)
+    all_positive = True
+    for token in profit_calcs:
+        if token[0] != "!":
+            if profit_calcs[token] < 0:
+                all_positive = False
+    profit_graph_data = json.dumps(profit_graph)
+    profit_calcs_data = json.dumps(profit_calcs)
+    spamwriter.writerow([txhash, drawn, unknown, all_positive, profit_calcs.get('ETH', 0), profit_graph_data, profit_calcs_data])

calculate_slots.py

@@ -0,0 +1,34 @@
+from statistics import mean
+import csv
+import numpy as np
+
+import csv_hack
+
+arbitrageurs = {}
+slotprices = {}
+
+def add_to_count(arbitrageurs, arbitrageur):
+    if arbitrageur in arbitrageurs:
+        arbitrageurs[arbitrageur] += 1
+    else:
+        arbitrageurs[arbitrageur] = 1
+
+slotsdict = csv.DictReader(open('data/gas_slots_6207336_6146507.csv'))
+slotsdict = csv.DictReader(open('data/gas_slots.csv'))
+for tx in slotsdict:
+    slot = int(tx['tx_position'])
+    if int(tx['gas_used']) < (int(tx['gas_limit']) * 0.6) and int(tx['gas_price']) > 310000000000 and tx['log_topics'].count("~") > 1 and not tx['to'].lower() in ["0xa62142888aba8370742be823c1782d17a0389da1", "0xdd9fd6b6f8f7ea932997992bbe67eabb3e316f3c"]:
+        print(tx['hash'], tx['from'], tx['to'])
+        add_to_count(arbitrageurs, tx['from'])
+    if not slot in slotprices:
+        slotprices[slot] = []
+    slotprices[slot].append(int(tx['gas_price']))
+
+for arbitrageur in arbitrageurs.keys():
+    if arbitrageurs[arbitrageur] > 0:
+        print("arber", "https://etherscan.io/address/" + arbitrageur, arbitrageurs[arbitrageur])
+
+open("data/slots_new.csv", "w").write("\n".join([",".join([str(x/(10**9)) for x in [ np.percentile(slotprices[slot], 10), np.percentile(slotprices[slot], 50), np.percentile(slotprices[slot], 75), np.percentile(slotprices[slot], 90), np.percentile(slotprices[slot], 99)]]) for slot in range(0, 10)]))
+for slot in slotprices:
+    prices = slotprices[slot]
+    print(slot, np.percentile(prices, 10), np.percentile(prices, 50), np.percentile(prices, 75), np.percentile(prices, 99))

count_wins.py

@@ -0,0 +1,23 @@
+import csv
+def get_winner_dict():
+    winner_dict = {}
+    slotsdict = csv.DictReader(open('data/slot_auction.csv'))
+    for slot in slotsdict:
+        slot['log_count'] = slot['log_addrs'].count("~") + min(1, len(slot['log_addrs']))
+        winner_dict[slot['hash']] = slot
+    return winner_dict
+
+
+arbs = {}
+
+winner_dict = get_winner_dict()
+print(len(winner_dict.keys()))
+
+for hash in winner_dict:
+    if winner_dict[hash]['log_count'] > 0:
+        sender = winner_dict[hash]['from']
+        if not sender in arbs:
+            arbs[sender] = 0
+        arbs[sender] += 1
+
+print(arbs)

csv_hack.py

@@ -0,0 +1,12 @@
+# QUICK AND DIRTY HACK FROM https://stackoverflow.com/questions/15063936/csv-error-field-larger-than-field-limit-131072
+import sys, csv
+maxInt = sys.maxsize
+
+decrement = False
+try:
+    csv.field_size_limit(maxInt)
+except OverflowError:
+    maxInt = int(maxInt/10)
+    decrement = True
+# END HACK
+

csv_to_sqlite.py

@@ -0,0 +1,4 @@
+import os
+
+os.system('sqlite3 data/arbitrage_new.db ".mode csv" ".import data/block_fees.csv block_fees" ".import data/eth.csv eth_data" ".import data/all_logs_bigquery.csv logs" ".import data/block_data.csv blocks" ".import data/all_success_arb_txs_bigquery.csv success" ".import data/all_inclfail_arb_txs_bigquery.csv wfail" ".import data/auctions.csv auctions" ".import data/profits.csv profits" "CREATE TABLE mergedprofitabletxs AS SELECT *,substr(timestamp,0,11) as date FROM wfail LEFT JOIN profits on profits.txhash=wfail.transaction_hash LEFT JOIN success on success.transaction_hash=wfail.transaction_hash LEFT JOIN blocks on blocks.number=wfail.block_number GROUP BY wfail.transaction_hash ORDER BY CAST(wfail.block_number as INTEGER) DESC, CAST(wfail.transaction_index AS INTEGER) ASC;" ".quit" ""')
+os.system('mv data/arbitrage_new.db data/arbitrage.db')

data/README.md

@@ -0,0 +1 @@
+Coming soon; check/star this space over the next day!

etherdelta/readme.txt

@@ -0,0 +1,33 @@
+The file contract_addr.txt contains all the contracts (including EtherDelta contract itself) with successful trade transactions sent to it within [3900000, 5550000) 
+
+for each file:
+
+all_txs/all_txs-{begin_block}-{end_block}-1.txt
+
+contains all the transactions sent to the addresses in contract_addr.txt in the block range [begin_block, end_block).
+
+There are 5 coloumns in each file, each line represents one transaction:
+
+BlockNumber TransactionHash From To GasPrice(Wei) GasUsed InputData
+
+
+for each file:
+
+succ_txs/succ_txs-{begin_block}-{end_block}-1.txt
+
+contains all the transactions with one or more Etherdelta Trade Event in the block range [begin_block, end_block),
+
+There are 6 coloumns in each file, each line represents one transaction:
+
+BlockNumber TransactionHash Tag From To InputData
+
+
+The Tag is one of { Trade, Arbitrage, Unknown}:
+
+Trade means this transctions only generated one Trade Event, which means it is a normal trade transaction.
+
+Arbitrage mean this transction generated exactly 2 Trade Events and the buy/sell tokens form a pair.
+
+Otherwise, the transation will be tagged as Unknown.
+
+

etherdelta/scripts/find_succ_txs.py

@@ -0,0 +1,71 @@
+import argparse
+from _pysha3 import keccak_256
+from web3 import Web3, HTTPProvider
+import json
+
+web3 = Web3(HTTPProvider('http://localhost:8549'))
+#web3 = Web3(HTTPProvider('https://mainnet.infura.io/Ky03pelFIxoZdAUsr82w'))
+
+etherDeltaAddress = '0x8d12A197cB00D4747a1fe03395095ce2A5CC6819'
+etherAddress = '0000000000000000000000000000000000000000000000000000000000000000'
+
+tradeAPI = '0x' + \
+           keccak_256(
+               b'Trade(address,uint256,address,uint256,address,address)'
+           ).hexdigest()
+
+parser = argparse.ArgumentParser(description='EtherDelta Arbitrage Bot.')
+parser.add_argument('--st',dest='st' ,type=int, action='store', default='5000000')
+parser.add_argument('--len',dest='len' ,type=int, action='store', default='100')
+parser.add_argument('--r',dest='r' ,type=int, action='store', default='20')
+args = parser.parse_args()
+
+startBlock = args.st
+endBlock = args.st + args.len
+ratio = args.r
+result_dir = '../results/succ_txs-{}-{}-{}.txt'.format(startBlock,endBlock,ratio)
+
+
+import os
+if os.path.isfile(result_dir):
+    print('Previous file exists.')
+    with open(result_dir, 'r') as f:
+        lines = f.readlines()
+        if(len(lines) >= 1):
+            print('last line:',lines[-1])
+            number = int(lines[-1].split()[0])
+
+else:
+    print('No previous file.')
+    number = 0
+
+
+
+for idx in range(max(startBlock, number), endBlock + 1, ratio):
+        block = web3.eth.getBlock(idx)
+        transactions = block['transactions']
+        print('block number:', idx)
+        for txHash in transactions:
+            tx = web3.eth.getTransaction(txHash)
+            receipt = web3.eth.getTransactionReceipt(txHash)
+            token_pair_list = []
+            for log in receipt['logs']:
+                if 'topics' in log and len(log['topics']):
+                    if log['topics'][0].hex() == tradeAPI:
+                      token_pair_list.append((log['data'][24 + 2: 64 + 2],log['data'][24 + 128 + 2: 64 + 128+ 2]))
+
+            num = len(token_pair_list)
+            tag = None
+            if num == 2 and token_pair_list[0][0] == token_pair_list[1][1] and token_pair_list[1][0] == token_pair_list[0][1]:
+                tag = 'Arbitrage'
+            elif num == 1:
+                tag = 'Trade'
+            elif num:
+                tag = 'Unknown'
+            if tag is not None:
+                result = "{} {} {} {} {} {}\n".format(idx, txHash.hex(), tag, tx['from'], tx['to'], tx['input'])
+                print(result)
+                with open(result_dir,'a') as f:
+                    f.write(result)
+
+

etherdelta/scripts/run_find_all_txs.sh

@@ -0,0 +1,19 @@
+#!/usr/bin/env bash
+
+onexit() {
+    kill -TERM -0
+    wait
+}
+trap onexit INT
+
+set -x
+
+for (( i=3900000; i<5550000; i+=10000))
+do
+    python3 find_all_txs.py --st $i --len 10000 --r 1 &
+
+done
+
+wait
+
+#python3 find_succ_txs.py --st 3900000 --len 100000 --r 20

etherdelta/scripts/run_find_succ_txs.sh

@@ -0,0 +1,21 @@
+#!/usr/bin/env bash
+
+onexit() {
+    kill -TERM -0
+    wait
+}
+trap onexit INT
+
+set -x
+
+for (( i=3900000; i<=5550000; i+=50000))
+do
+    python3 find_succ_txs.py --st $i --len 50000 --r 1 &
+#   cmd="python3 find_succ_txs.py --st $i --len 100000 --r 20 &"
+#   echo "running $cmd"
+#   eval $cmd
+done
+
+wait
+
+#python3 find_succ_txs.py --st 3900000 --len 100000 --r 20