Polymarket Bot Tutorial · Sura ya 14 kati ya 32
News arbitrage kwenye Polymarket: jinsi ya kushinda market kwenye headlines, source feeds (RSS/Twitter/AP), latency budgets, false-positive filters, na wakati news edge inakufa katika market price.
Sura hii inafunika nini
News arbitrage ni strategy ya kutrade public information haraka kuliko market inavyopanya tena. Edge ni real lakini narrow - "news" nyingi tayari iko katika price kufikia wakati binadamu anaweza kuisoma. Sura hii inafunika ni sources zipi zinashinda market, latency budget inayodefini strategy, na false-positive filter bila ambayo bot inatrade kwenye kila retweet.
- Information edge inaonekanaje
- News sources: RSS, Twitter, AP, official feeds
- Latency budget: read-to-trade chini ya sekunde 2
- False-positive filters
- Wakati news edge inakufa
- Code: poll news feed na weka FOK kwenye relevant markets
- Risk: half-truths na walked-back headlines
Information edge inaonekanaje
News arbitrage inamaanisha kutrade public information haraka kuliko market inavyopanya tena. Edge inakuwepo katika narrow window - kawaida sekunde 30-300 - kati ya ukweli kuwa public na Polymarket kureflect.
Ili edge iwe real, vitu vitatu lazima viwe kweli. Kwanza, news source lazima iwe haraka kuliko median Polymarket trader (Twitter ni haraka kuliko mainstream press; AP wire ni haraka kuliko Twitter). Pili, news lazima iwe unambiguous (injury announcement, court ruling) - interpretation inakula latency. Tatu, market lazima iwe wide ya kutosha kwamba price move iko worth spread tax.
Bots wanaohunt edge hii wanagawanyika katika camps mbili: zile zinazojisajili kwa direct sources na kupasse, na zile zinazoangalia unusual price move kwenye Polymarket na kuinfer news ilitokea. Zote ni valid; ya kwanza inaongoza, ya pili inafuata.
News sources: RSS, Twitter, AP, official feeds
Sources zilizopangwa kwa latency hadi public-information-status, haraka kwanza.
- Direct primary sources: court filings, government press releases, central-bank announcements. Mara nyingi zina public RSS au API. Haraka zaidi, false-positive rate ya chini zaidi.
- AP wire / Reuters Eikon (paid). Wire ambayo traditional traders wanatumia. ~Sekunde 5-30 lead kwenye consumer Twitter.
- Twitter (X, paid API). Lists za verified accounts: official org accounts, beat reporters. Free APIs ni rate-limited sana; lipa kwa Pro tier au tumia relay service.
- Specialized newsletters / Discord: paid Substacks, embargoed industry feeds. Inafaa kwa niche markets (crypto, esports).
- Mainstream press websites: polepole sana kwa news-arb edge.
RSS kwa kila kitu kinachopublish RSS - ni bure, polling intervals ni reliable. Twitter kwa rest. AP kwa production-serious news desks.
Latency budget: read-to-trade chini ya sekunde 2
Bot inahitaji kuingest, kuclassify, kuamua, na kuweka order ndani ya sekunde 1-2 jumla. Budget:
- Ingest: 50-300ms (websocket feed, RSS poll, Twitter stream).
- Classify: 50-200ms (regex / keyword match, optionally LLM ikiwa unacache prompt).
- Decide: 50ms (rules table lookup; mapping kutoka news tag hadi market slug).
- Place: 200-500ms (FOK signed order kwa CLOB).
Budget eater mkubwa zaidi mmoja ni LLM classification. 500-token GPT-4 call inaongeza sekunde 1-3; hiyo ni arb window nzima imeenda. Kwa production, classify na keyword rules; tumia LLM tu kwa offline calibration ya keyword set.
False-positive filters
News-arb bots ambazo hazifilter false positives zinatrade kwenye kila retweet na zinableed kupitia spread tax. Filters tatu.
- Source whitelist: act tu kwenye accounts/feeds katika pre-approved list. List ni ndogo (sources 10-30).
- Keyword + confirmation pair: single keyword match ni noise; matches katika sources mbili huru ndani ya sekunde 30 ni signal.
- Market-state guard: skip markets ambazo tayari zimehama > 5% katika sekunde 60 za mwisho - mtu mwingine alikamata news kwanza, edge imeenda.
False-positive rate ya well-tuned filters: takriban 1 katika 5-10. 90% false-positive rate inaharibu strategy; 50% rate ni workable na small position sizes.
Wakati news edge inakufa
Window kutoka "news public" hadi "price reflects news" inafunga haraka zaidi kila mwaka. Mnamo 2020, mid-priced political markets zilichukua dakika kuabsorb headline. Mnamo 2026, headlines sawa zinacompress hadi sekunde 30-90 kabla price imehama kabisa.
Signs edge imekufa: per-trade PnL kwenye flagged trades inadrop kutoka +3c hadi flat juu ya window ya trades 30; rate ya false positives zinazogeuka kuwa already-priced-in inapanda juu ya 70%; market inahit FOK ask yako ndani ya 200ms kwa sababu mtu mwingine alifika kwanza.
Honest pivot wakati edge inakufa: hamia kwa slower, more interpretive news (court rulings, central bank meeting minutes) ambapo kuparse meaning kunachukua muda mrefu zaidi kuliko latency race. Au acha kuendesha strategy.
Code: poll news feed na weka FOK kwenye relevant markets
Production skeleton: poll news source, run rule matches, fire FOK orders kwenye hits.
import feedparser, time, re
from py_clob_client.client import ClobClient
RULES = [
{"regex": re.compile(r"out for season|torn ACL", re.I), "tag":"injury-fade"},
{"regex": re.compile(r"federal reserve.*(rate cut|rate hike)", re.I), "tag":"fed-move"},
]
seen = set()
while True:
feed = feedparser.parse("https://example.com/news.rss")
for entry in feed.entries[:20]:
if entry.id in seen: continue
seen.add(entry.id)
for rule in RULES:
if rule["regex"].search(entry.title + " " + entry.summary):
# Look up relevant Polymarket markets, place FOK
fire(rule["tag"], entry)
break
time.sleep(15)Polling intervals: sekunde 5-15 kwa RSS. WebSocket ambapo inapatikana (Twitter, AP wire). Daima dedup kwa source-provided ID; kamwe usidhani polling ni exactly-once.
Risk: half-truths na walked-back headlines
Worst day ya news-arb bot ni wakati headline inageuka kuwa wrong. Mifano: Reuters tweet inasema "Trump fires Yellen," market inaruka 8 cents, dakika 12 baadaye tweet imefutwa na imecorrected. Bot iliyonunua kwa +8c sasa inashikilia inventory kwa -3c bila recourse.
Defenses:
- Two-source confirmation: kamwe usitrade kwa single tweet; require corroborating signal kutoka kwa second independent source ndani ya sekunde 60-180.
- Position size scaled kwa source confidence: AP wire = full size; Twitter kutoka kwa verified beat reporter = 50%; rumor source = 25%.
- Auto-exit kwenye retraction signal: ikiwa source uliyotumia inaitoa correction ndani ya dakika 30, exit kwenye market bila kujali PnL.
Walk-back problem ni hard ceiling kwenye news-arb position sizing. Kutrade $50 per signal kunakuruhusu kuishi 30% false-positive rate; kutrade $500 hakuruhusu.
