Where edges come from.
An edge is never a feeling. It is a specific piece of public information that the price has not absorbed yet, which means it always has an address: a team sheet, a filing, a polling average, a funding history, a weather ensemble. Here is where each family keeps its evidence, and how long that evidence stays worth anything.
Edges come from primary sources read early, not from commentary. Every family has its own: line-ups and injury reports in sport, polling aggregates and filings in politics, release calendars in macro, funding history and on-chain data in crypto, forecast ensembles in weather, trade press in culture, public counters online. The value of each decays fast, which is why the strategy that uses them has to run every day.
01Primary beats commentary, every time
By the time an article explains why a price moved, the price has moved. The sources that produce edges are the ones the article was written from: the official team sheet, the actual filing, the raw poll, the number on the exchange. The skill is not access, since almost all of it is public and free. It is the habit of going one step upstream of everyone else.
02Sport: the team sheet is the document
Line-ups published around an hour before kickoff, injury and suspension lists, rotation in a congested week, travel and rest days, and in some competitions the referee appointment. In the biggest fixtures this is priced within minutes. In the twelfth match of a busy midweek, it can sit unpriced for an hour, and that gap is the entire opportunity. Decay: minutes in a headline match, hours in a quiet one.
03Politics: aggregates, not single polls
Polling averages rather than the poll that made the news, filing deadlines and candidate registrations, official announcements, and, above all, the resolution source the market itself names. A single startling poll is noise that the average will absorb. The market that says it settles on a specific certification is a different bet from the one that settles on a call by a network. Decay: days, sometimes weeks, since these markets reprice slowly.
04Macro: the calendar is half the work
Central bank meeting dates, inflation and jobs release times, the published consensus, and the wording of the threshold in the market. Knowing that a print lands at half past one and that the consensus sits at a particular figure tells you when the repricing happens and what would surprise it. Decay: seconds at the release, then a slower second wave in thinner related markets.
05Crypto: the venue publishes your data
Funding history on perpetuals, open interest, listing announcements, protocol upgrade schedules, on-chain flows. This is the family where the raw material is most generously published and least often turned into anything. A funding series read as a time series rather than a snapshot tells you when one side of a market has become crowded enough to pay for the other. Decay: hours.
06Weather: probabilistic sources for probabilistic questions
National meteorological services publish ensemble forecasts, which are already probabilities, and historical station records give the base rate. That makes this the cleanest comparison on the whole board: a model probability against a market price, with a public measurement settling it. Decay: a forecast horizon, so days rather than minutes.
07Culture: precursors and the trade press
Guild and precursor awards predict the big ceremonies better than opinion does. Trade publications carry casting and release news before the general press. Box office tracking is published days ahead of a weekend. Decay: slow, since these crowds react to sentiment rather than to schedules.
08The internet: the counter is the source
Post counts, view counts, follower milestones and shipping dates all resolve on a public number anyone can read. The edge is almost entirely base rate: what did this account average over the last ten weeks, what does the current pace imply, how does the weekend change it. Unglamorous arithmetic, done before anyone else bothers. Decay: continuous, because the counter keeps moving.
09The most underrated source: the market itself
The rules page names the resolver, the deadline and the edge cases, and it is where the mispricing frequently hides. Related markets are a source too: a contradiction between two prices is evidence one of them is wrong, and it costs nothing to check. So is the order book, which shows whether an edge is large enough to survive getting in and out.
10Sources decay, which is the whole problem
Every item above stops being an edge the moment enough people read it. That is why a single good source is worth less than a routine that checks many of them, on a rhythm, without getting tired or bored or attached. A person can do this for one family for a few weeks. Sustaining it across twelve is a different kind of job.
A source you cannot name is not a source. If the answer to "why would the crowd be wrong here" is a feeling, the crowd is probably right and the fee is certain.
11How your CEO does it
Your CEO writes its own instruments to read these sources, stores what it finds as its own data, and compares its number against the market before proposing anything. It cites what it read when it explains a bet, so you can ask where a figure came from and get an answer in plain words. You can also point it: tell it to watch the weather desk in summer, to stay out of politics, or to go deep on one league. The sources are public. Reading all of them, every day, is the part you are buying.
What to remember
- Go upstream. The document the article was written from is the source.
- Each family has its own evidence, and its own decay clock, from seconds to weeks.
- The rules page and the order book are sources most traders never open.
- Coverage beats cleverness: one source read late is worth less than many read early.
- If you cannot name the source, there is no edge, only a fee.