Brand
Logo, colors, voice
Drop these into your video edits, banners, and graphics. Hex codes are exact.
App icon β 192px
/icons/icon-192.png
App icon β 512px
/icons/icon-512.png
Voice guide: Direct, numbers-first, never hype. "We tell you, fast, when our model and the market disagree, and we publish every call" is on-brand. "We tell you when the market is wrong" is not, and neither is "GUARANTEED LOCK πππ". Always pair a win-rate with a loss-clause, using the LIVE number for the product you cite β e.g. "79.0% over 852 MLB games at a median entry of 80Β’, close to breakeven before fees β about 1 in 5 still loses."
Sports Β· Market 1 of 4 Β· Live Β· Record still empty
π NFL In-Game Signals
Live NFL win-probability vs. live Kalshi KXNFLGAME prices, on Telegram. Running now, and the public record is still empty: no alert has fired yet. Customer needs a Kalshi account.
β οΈ One NFL number is citable, it is a backtest, and it loses money.
The live record is zero and the log is still empty. Not one NFL alert has fired, so there is no NFL live win rate, no entry price, no streak and no sample size. Say it plainly: "NFL has no published track record. It is still empty: Week 1's opener produced no alert at all. The public NFL record starts with the first alert that fires, and none has yet β and when one does, it goes into the log, losses included." Post the public results log so people can watch it fill in. Nothing you post may date the record to kickoff or to a named day, or call it live and filling in: kickoff has passed, the log is still empty, and every one of those constructions is now false.
The one citable cell, re-cut 2026-09-10. The model gate shipped from 0.80 to 0.70 today, so the cell that describes the product changed with it. Paste this paragraph whole or leave the number out β there is no short version: "In a 2025-26 season backtest, at the same gate the live bot now ships with, entering two minutes after detection the way the bot really runs, the model won 60.9% of 225 graded signals at a 62.5 cent average entry. A 62.5 cent entry has to win 62.5% of the time to break even, so the backtest finishes 1.6 points below breakeven before Kalshi fees, and further below after them. It is not a winning backtest. The delay-two grid has sixteen cells, it is negative on average, and its one positive cell is the model-0.80 gate the bot just left, so the cell that describes today's product is one of the fifteen negative ones and was not picked for looking good. The public NFL record starts with the first alert that fires, and none has yet." The full replay, the interval, the delay decay and the EV arithmetic are on the research page; link it wherever the format allows a link, and leave the EV number there.
You may: quote that cell only in a format with room for the whole paragraph above β the word backtest in the same sentence, "60.9% of 225 graded signals" with the 62.5Β’ average entry beside it, the explicit subtraction that puts it 1.6 points below the breakeven that entry implies before fees, the sixteen-cell sentence with both of its halves, and "the public NFL record starts with the first alert that fires, and none has yet." A tweet, a one-line DM and a spoken hook do not have that room, so in those formats link the research page and quote no number at all. Never present it as a winning number: it is not one, before fees or after.
You may not: drop the word backtest, the entry price, the graded denominator or the below-breakeven subtraction β the shortfall is not optional in any format, because it is the harder statement and it is the one we make; let the fired count stand in as the denominator β the split is "228 fired, 225 graded, 137-88", and 225 is the only denominator that survives beside 137-88; present it as a live record, a streak, or "our results"; date the start of the record to kickoff or to any named day; round or restyle it ("nearly 61%", "6 in 10", "about 61%"); drop either half of the sixteen-cell sentence; quote any other cell, gate or delay from that study, and in particular never the retired model-0.80 row, its win rate or its entry price; state a margin, an EV number, or any profit, income or "+EV" claim, in copy or in a footnote β a footnote is still copy, and the EV number stays on the research page; blend it with the MLB, NBA or BTC numbers; or put an MLB or BTC figure beside the NFL pitch where it reads as NFL performance.
Where you have room for a footnote, add the caveat: one season is a thin sample; measured per game rather than per signal, the same cell wins 62.2% of 119 games with a 95% interval running 53.2% to 70.4%. Do not subtract the 62.5Β’ entry from that game rate β the entry price is the average across the 225 graded signals, a different population, and the difference between the two is not a margin. The one thing that is not wrong with this cell is selection: we ship it because it is the gate the bot runs, not because it looked good, and it did not. A footnote is copy like any other, so it carries the same bans β no EV number, no margin, no profit claim. Every post carrying your code, your link or comped access also carries an FTC disclosure: see the compliance section.
0
NFL alerts published β the record starts with the first one
None
Live NFL win rate, entry price, streak or sample
$4.99
Per month, after a 30-day free trial
32
Teams covered Β· /teams filter
The whole offer, state it in full: free for 30 days, card required at checkout, $0 today, cancel before day 30 and you pay nothing, then $4.99/mo. Never shorten this to "free" β a card is required and it renews.
Your NFL link: https://buy.stripe.com/14AaEX8GBeVyfDsapO1Nu0p β use {{REF_URL}} in the templates below and the tool swaps in your own tracked version of it.
π¦ Tweet templates
DegenHedge NFL in-game alerts are running.
Model win probability vs the live Kalshi price. When the two disagree past the threshold, you get a Telegram alert. We do not claim the price then moves toward us.
No published track record, and the log is still empty β Week 1's opener produced no alert at all. It starts with the first alert that fires, losses included.
Free for 30 days, card required, then $4.99/mo. {{REF_URL}}
A live win-probability model and a live Kalshi price are two different numbers. DegenHedge measures the distance between them and pings your Telegram when it is wide.
What it is not: a claim the price moves after the alert. We instrumented that and it does not. What you get is the detection, fast, and a public log graded at settlement.
Zero NFL alerts published so far β Week 1's opener produced none, so the log is still empty and it starts with the first one that fires. Watch it fill in instead of taking my word for it: degenhedge.com/results/
Free 30 days, card required, then $4.99/mo. {{REF_URL}}
The NFL feed is running with no published record. Not one alert has fired yet; the record starts with the one that does.
What exists is a season replay, and it finishes below the breakeven its own entry prices imply. We published it anyway, with the math open. It does not compress into a tweet without losing the part that matters, so here it is whole:
degenhedge.com/research/kalshi-nfl-repricing-lag/ β read it before you buy anything.
Free 30 days, card required, then $4.99/mo. {{REF_URL}}
π¬ DM / Discord pitch
Quick one β DegenHedge NFL in-game alerts are running. It runs a live win-probability model against Kalshi's KXNFLGAME prices and fires a Telegram alert when the two disagree. It does not claim the price then moves toward the model.
Being straight with you: NFL has no published track record, and it is still empty β Week 1's opener produced no alert at all. The public record starts with the first alert that fires, and none has yet. Every one gets written to their public results log as it fires, losses included, so you can judge it in real time rather than on a claim.
Free for 30 days, card required at checkout, $0 today, cancel before day 30 and you pay nothing, then $4.99/mo. My link: {{REF_URL}}
DegenHedge NFL in-game alerts are running. Live win-probability model against Kalshi's KXNFLGAME prices, Telegram alert when the two disagree. No claim that the price then moves toward the model.
No published track record, and it is still empty: Week 1's opener produced no alert at all. The public record starts with the first alert that fires, and none has yet.
The one number that exists is a backtest of last season, at the gate the bot now ships with, entering two minutes after detection the way the bot really runs: 60.9% of 225 graded signals at a 62.5 cent average entry. Read those two together, because a 62.5 cent entry has to win 62.5% of the time to break even β the backtest finishes 1.6 points under that before Kalshi fees and further under after them. It is a losing backtest and I am not selling it to you as a winning one. The delay-two grid has sixteen cells, it is negative on average, and its one positive cell is the model-0.80 gate the bot just left, so the cell that describes today's product is one of the fifteen negative ones and was not picked for looking good. One season is a thin sample. The whole replay is public: degenhedge.com/research/kalshi-nfl-repricing-lag/
Free for 30 days, card required at checkout, $0 today, cancel before day 30 and you pay nothing, then $4.99/mo. Analysis for entertainment; any single alert can lose 100% of the stake. My link: {{REF_URL}}
π¬ Video script bullets
HOOK (0-3s): "NFL alerts in my Telegram during the game. Here's the mechanism, and here's what I don't know yet." (No number in the hook. The one NFL figure that exists is a losing backtest and it cannot be said in three seconds with the caveat it has to carry.)
PROBLEM (3-20s): You are watching a game and a live market at the same time, and you cannot do the arithmetic between them in your head fast enough to matter.
THE MECHANISM (20-45s): Model reads score, clock and situation, prices the game, compares to the live Kalshi contract. The two disagree, the alert fires, the call goes into a public log. Say out loud: we do not claim the price moves toward us afterward, and our own order-book data says it does not.
THE HONEST PART (45-65s): "There is no NFL win rate. Not one graded alert β the opener produced none, so the public log is still empty and it starts with the first call that fires, losses included β degenhedge.com/results/. Judge it from there, not from me. There is a season backtest, and it finishes below the breakeven its entry prices imply. It's linked below." (Put the link in the description. If you quote its number at all, paste the whole paragraph from the asset kit's NFL callout into the description β never the headline number on its own, and never an EV figure.)
CTA (65-90s): "Free for 30 days, a card is required and it renews at $4.99/mo, cancel before day 30 and you pay nothing. Link in the description. Size so a losing streak would not bust you. Read the risk page."
Sports Β· Market 2 of 4 Β· In Season
βΎ MLB In-Game Signals
Live MLB win-probability via empirical run-differential table. ~7-9 alerts per game-night across the league. Customer needs a Kalshi account.
79.0%
Win Rate (V2.2, median entry 80Β’ β close to breakeven before fees)
1 in 5
Signals that lose 100% (about 1 in 5)
π¦ Tweet template
MLB season is underway and DegenHedge has logged 79.0% WR on in-game signals at a median entry of 80Β’, which is close to breakeven before fees.
Empirical 2,232-state win-prob table vs live Kalshi prices. Model fires when the diamond and the market disagree.
~7-9 signals/night across the league. {{REF_URL}}
Got the alert at [INNING]: [TEAM] at [PRICE], model said [MODEL PROBABILITY].
Win.
This is what DegenHedge does: it flags the gap between its model and the Kalshi price during games. 79.0% WR live at a median entry of 80Β’, which is close to breakeven before fees. Any alert can lose 100%. {{REF_URL}}
π¬ DM / Discord pitch
If you bet MLB live β DegenHedge fires Telegram alerts during the game when the model sees a gap vs Kalshi. Live WR 79.0% over 852 games at a median entry of 80Β’, which is close to breakeven before fees. Filter by your teams in the bot. Free for 30 days, then $4.99/mo. My link: {{REF_URL}}
π¬ Video script bullets
HOOK (0-3s): "Telegram MLB alerts, 79.0% winners over 852 graded games at a median entry of 80Β’, which is close to breakeven before fees."
CONTEXT (3-20s): MLB has 2,430 games a season. Way too many to watch live. But Kalshi prices them all.
THE MODEL (20-45s): 2,232 game states β inning, outs, run differential, half-inning. The model knows what each state historically meant. When Kalshi disagrees, alert fires.
PROOF (45-65s): degenhedge.com/mlb β 852 games, 79.0% WR at a median entry of 80Β’, which is close to breakeven before fees. Refreshes every 30 seconds.
CTA (65-90s): "Filter to just your teams. Free for 30 days with my link, link in the show notes. Size so a losing streak would not bust you. Risk page in the show notes."
Crypto / Prediction Markets Β· Market 3 of 4
βΏ Kalshi BTC 15-Minute Signals
BTC alerts for markets on Kalshi, a CFTC-regulated US exchange. DegenHedge is a publisher, not a regulated entity. Customer needs a Kalshi account.
β οΈ Never pitch BTC as an edge or an opportunity.
Do the price math first. These alerts fire at a 91.82Β’ average entry (96Β’ median), and at a 91.82Β’ entry you have to be right 91.82% of the time just to get your money back. The logged rate is 91.3% across 9,750 priced signals, which sits essentially at that breakeven before fees, and below it after them. A high win rate on contracts that already cost almost a dollar is not an edge, and quoting the rate without the price turns it into one.
You may: describe what the feed does (a fast alert about 2 minutes before a 15-minute window resolves), quote 91.3% only with "across 9,750 priced signals at a 91.82Β’ average entry, 96Β’ median" in the same sentence, and say out loud that wins are small and a single loss costs nearly the full stake.
You may not: call it an edge, a mispricing you can profit from, "free money", "the cleanest edge on Kalshi", or any variant that implies the rate beats the price. No margin, EV, profit or income claim, ever.
Also banned everywhere, all markets: that the line is slow, that the book has not caught up, that the market has not repriced, that we time or catch a lag, or that the gap closes. We instrumented our own order book and the alerted side does not measurably move β median 0.0000¢ out to five minutes, flat spread, rising depth. Pitch the detection and the public log, never a price move. Retired figures, banned in every channel: every NFL figure this pack carried before 10 September 2026 β the model-0.80 gate the bot no longer runs, in every form, including its win rate, its graded count, its entry price and its EV; and the older BTC pair, a higher headline rate and a larger signal count that counted rows carrying no price at all. We do not restate the digits here, because a retired number outlives the sentence attached to it. If an old draft or a post still carries one, replace it with the re-cut NFL cell in the NFL callout above or the corrected BTC line above, the same day, without repeating the old number. Retired dates go with them: anything that dates the start of the NFL record to kickoff or to a named day, or calls it live and filling in, is now false β kickoff has passed and the log is still empty.
91.3%
Win Rate (alert-eligible, 91.82Β’ average entry, 96Β’ median)
~2 min
Alert lead time before each window resolves
10,129
Priced Signals Logged
91.82%
Break-even at the 91.82Β’ average entry (the logged rate sits at it, not above it)
π¦ Tweet template
Kalshi BTC 15-min signals have graded 91.3% over 10,129 priced alerts, at a 91.82Β’ average entry (96Β’ median).
Read those together: at 91.82Β’ you need 91.82% just to break even, so that rate is level with the line before fees and under it after them. It buys you speed, not an edge.
Telegram alert ~2 min before each window resolves. Free for 30 days with my link: {{REF_URL}}
How the BTC 15-min feed actually works, and why I do not call it an edge.
In the last 60 seconds before a window resolves, the model forms its own read of the likely outcome from the price action, and the alert reports where that sits against the quoted contract. The catch is the price: these alerts average a 91.82Β’ entry, where you need to be right 91.82% of the time just to break even, and the logged rate is 91.3% across 10,129 priced signals, 96Β’ median. Small wins, and a loss costs almost the whole stake.
It is an information and entertainment tool. Any alert can lose 100%. 18+. 1-800-GAMBLER.
Telegram alerts via @DegenHedge: {{REF_URL}}
π¬ DM / Discord pitch
Hey β saw you trade prediction markets. There's a Telegram alert service called DegenHedge that flags BTC 15-min windows on Kalshi where its model disagrees with the market, about 2 min before each window resolves. Logged record is 91.3% over 10,129 priced signals, at a 91.82Β’ average entry (96Β’ median). Breakeven at 91.82Β’ is 91.82%, so that rate sits level with the line before fees and under it after: wins are small, a loss costs almost the full dollar, and I am not selling it to you as an edge. Not a bot trading β just alerts to your phone, you place the bet.
Free for 30 days with my link: {{REF_URL}}
π¬ Short-form video script bullets (60-90 sec)
HOOK (0-3s): "Kalshi's 15-minute Bitcoin windows have a very high hit rate on contracts that already cost most of a dollar. Do the breakeven math before you act on one." (Keep the rate out of the hook β it only means anything next to its entry price and its sample, and those do not fit in three seconds. Both are in the PROOF beat below.)
PROBLEM (3-15s): With ~60s left in a window, a model reading the price action forms a view. The quoted contract is a separate number. Say plainly: this is a disagreement, not a claim that the quote is about to move.
PROOF (15-40s): Show the live track record at degenhedge.com β 91.3% across 10,129 priced signals at a 91.82Β’ average entry, 96Β’ median, which is level with breakeven before fees and under it after.
DEMO (40-70s): Show a real alert from the expensive end of the range: "UP, model 99%, market $0.98. At 98Β’ the contract has to win 98% of the time just to return the stake before fees, and across the whole log the hit rate does not clear the break-even implied by its own average entry." Walk through where the price sits on Kalshi, and say out loud that this is not presented as an edge.
CTA (70-90s): "Free for 30 days with my link. Link in bio. Bet small; every signal can lose 100% of the stake. Disclaimer in pinned comment."
Sports Β· Market 4 of 4 Β· Waitlist only
π NBA In-Game Signals
Offseason. NBA alerts, NBA checkout and the public NBA record all open at tip-off in late October.
β Do not quote any NBA number.
There is no publishable NBA win rate. The NBA log covers one partial season and the entry prices behind it are being re-derived from the live pipeline, so we do not publish a rate and neither may you. That includes the win rate that used to sit in this pack: it is retired, and any post still carrying an NBA number should be edited or taken down. Terms Β§ 2.3(f) and Appendix A ΒΆ A-6 make this a flat ban, not a caveat.
The only permitted statement: "NBA alerts open at tip-off in late October." Point people at the waitlist, not at a checkout β the NBA payment links are inactive until tip-off, so a link you post now cannot convert and cannot pay you a bounty.
When NBA goes on sale, this section gets a full kit on the same terms as every other sport: free for 30 days, card required at checkout, $0 today, cancel before day 30 and you pay nothing, then $4.99/mo. Until then, promote MLB, NFL or BTC.
Multi-sport
π Multi-sport pitches
There is no bundle. Each DegenHedge sport is $4.99/mo on its own, free for the first 30 days with a card required, so a follower who wants two sports starts two trials. One trial per person per sport.
π¦ Tweet β every feed
DegenHedge runs in-game feeds on Kalshi: MLB, BTC 15-min and NFL are all running today. NBA is back at tip-off.
Where the record stands, with the price you'd have paid:
β’ NFL: no published record, and still empty. It starts with the first alert that fires, and none has yet.
β’ MLB: 79.0% over 852 games at a median entry of 80Β’, close to breakeven before fees
β’ BTC 15-min: 91.3% across 10,129 priced signals at a 91.82Β’ average entry, 96Β’ median β level with breakeven before fees
Each DegenHedge sport is $4.99/mo on its own, free for the first 30 days with a card. {{REF_URL}}
π¦ Tweet β sports only
If you only bet sports: DegenHedge NFL in-game alerts are running and so is the MLB feed. NBA comes back at tip-off. Each DegenHedge sport is $4.99/mo on its own, free for the first 30 days with a card, and you never pay for the BTC alerts you don't use.
MLB: 79.0% over 852 games at a median entry of 80Β’, close to breakeven before fees. NFL: no published record, still empty β it starts with the first alert that fires, and none has yet.
{{REF_URL}}
FAQ for affiliates
Stuff you'll wonder about
New terms. The program now pays a flat $10 bounty per converted referral instead of a revenue share. The answers below are the current terms. Creators on written percentage terms from before the change stay on them until they elect the bounty in writing.
When do I get paid?
Monthly, on or about the 15th, via Stripe Connect or wire, covering bounties that became payable in the prior month. A bounty becomes payable 30 days after the conversion. Minimum payout threshold is $25 and anything under it rolls into the next month.
How much is the bounty?
A flat $10 for every referred person who converts from the free trial to a paid subscription, the same on every tier. It is paid once per person, not once per sport, so ten conversions is $100.
What if a customer cancels and re-subscribes?
Nothing further is paid. At most one bounty is ever payable for a given person, whether they cancel and come back, add another sport, or check out again under a different creator's code. The first affiliate attributed to that person earns it.
Can I run paid ads with my code?
Yes, with two rules: (1) you bear all the ad spend; (2) you cannot bid on the keyword "DegenHedge" or any close variants in Google / Bing / X β that competes with our direct traffic. Native social ads, Discord pitches, podcast reads are all fine.
What happens if my audience already had a referral cookie from someone else?
Last-click wins. If they visit your URL, your code overwrites the previous one in their browser (the localStorage entry resets). When they pay, you get the credit.
Can I get a custom code?
Yes β when you applied, you saw a "code" field. If you didn't fill it, your code was generated from your name. Email
[email protected] to request a swap.
Can I see real-time stats on my referrals?
Today: not yet β email
[email protected] and we will share your stats. A self-serve dashboard at degenhedge.com/affiliate-dashboard is on the roadmap.
My audience asks why this isn't auto-trading. What do I tell them?
Because market makers on Kalshi don't fill big automated orders near resolution, so the alerts are built around human-scale bets ($5-100). Anything larger moves the book against itself. Send them to
degenhedge.com/method/#lag for the long answer.
Can I see which of my followers actually subscribed?
No β privacy reasons. You see counts (referrals, lifetime owed), not identities. Email
[email protected] if you have a specific dispute.