General discussion about Bifrost goes here.
If you have bugs or need support, make a post here: Support - Valhalla
If you’d like to request a feature, make a post here: Feature Requests - Valhalla
Otherwise, discuss it here.
1 Like
End Of Week Wrap-Up
First week of Bifrost is in the books and it’s been good to be back in the room with some of you. Truth is we watched a lot of winners this week instead of trading them. That changes. It changes as everyone gets their feet wet with the system and starts thinking through trades as far as Heimdall pushes you to.
That’s not to say we didn’t trade winners because of course we did. OKLO was 100% gain today, OPEN was 400% yesterday. Both simply called by Heimdall and were green holds almost the whole way.
One thing that’s easy to gloss over is that Heimdall is as much about limiting losses as finding plays. It gives you a framework to operate inside and tells you, in plain terms, where the play is invalidated. Like I said in VC, part of why I went this direction is to get us out of the business of “calling the shot.” When you call it, you own it, you get attached to it, and attachment is where the bad decisions come from. It’s hard to be objective about a play you called. It’s easy to be objective about one you didn’t. With Heimdall, nobody in this room is spending the day on the heavy research. We’re ingesting it and filtering it together.
As you have time this weekend, think through what you’ve seen and open tickets. Features you think are missing go in Feature Requests - Valhalla, bugs go in Support - Valhalla.
Foot On The Gas
The server is still waking up. Development is not. Over the weekend:
- The first major model update for Heimdall. Fixes several known issues we found this week and makes a substantial upgrade to the model’s capability.
- Institutional news sources. I acquired access today to a feed carrying analyst upgrades and downgrades, CNBC, Reuters, and two private institutional newswires. It also carries real ticker fields for PRNewswire and Globe Newswire, so we’re no longer parsing tickers out of press releases in code. Better news means better arcs, and it adds a key missing component: Heimdall waking up the moment an analyst rating changes around a catalyst like earnings.
- Bugs and plumbing. Several of the open bugs, plus large changes to the Positions integration, Arc comments, (attachments for flywoman) and more.
This Is Only The Beginning
Go look at the Heimdall Hall of Fame again: Heimdall Hall of Fame
The goal right now is simple: get the people I built this for into the room while I tune it against real usage. Bifrost was built to have eyes on it. Catalysts worked best when the room was full and everyone was contributing, and that is where I’m driving this. If you’re still in touch with someone who left, or who doesn’t know they have access, send them this. It’s been years in the making. Don’t let them miss it. Server is wide open, I don’t care what was said, how salty they are, take the invite and come look at this. It’s right here: https://discord.gg/N6BwMUq89
And After That
Soon I’ll be trading alongside the community again as we open the gates, for the first time in years, to a wider audience. Bifrost will bring new people in, and we’ll build something stronger and more profitable than any iteration of this community has achieved before. Membership stays capped, like I said. I have no interest in building a second WSB. I’m building Renaissance Technologies for a select few. A Viking roleplay version of the Freemasons.
The vision is playing out, and we’ll look back on these weeks fondly. I’m not building this community back to what it was. I’m building it to what it will be. Afterhours will be a celebration again. Trading-floor is already full of green again, BUT THIS TIME IT DOESN’T GO AWAY. You can already see it: three of us traded today the way it used to take a full server to do. IMAGINE what this looks like with the room full. THIS IS THE VISION. This is the last iteration of Valhalla, because this one doesn’t die. It’s the mic drop. The next big play retail finds is going to pump positions we ALREADY HAVE, because we’re on a different level now. Get active, get commenting, get reading.
Every challenge I’ve ever run finished with a win rate above 80%. Nobody takes this more seriously than I do. That’s what the year and a half was for. So when I say it again, the right bet is that I’m telling you the truth: GREATEST. TRADING. COMMUNITY. IN. EXISTENCE.
Oh and yeah, I’m thinkin’ I’m back. Rest up, we’re back at it Monday. 
1 Like
I ended up applying part of the Heimdall model update. The price action fix ended up being far more significant than I had originally thought and was a fairly large rework of some of Heimdall’s reasoning in some areas so in order to properly track how it’s affecting signals I’m running it for a couple days without the additional updates.
All the other changes listed were either rolled out this weekend or will be rolled out over the course of today and tomorrow.
Just a small announcement. Discord charts now support missing ETFs like VIX. They’re are now 24H on supported tickers, etc.
End Of Week Wrap-Up
Week two of Bifrost and, well, things are already heating up. Remember how I said last week that we watched a lot of winners and how that would change this week? Well, it changed.
Some Stats
Positions
This week 12 people used the positions system to log 67 trades based on Heimdall’s signals.
75% of those people were green.
Together they made $6,406.64 or over +600% on signal-linked trades.
They had an average single trade gain of 50.8%.
3 of those people are on a Free Trial.
They were profitable across 16 total signals, so 82.4% of signals they traded.
All 6 logged trades that had no Bifrost coverage were red.
Signals (IV Excluded)
Heimdall put out 190 signals across 169 tickers this week for an average of 24 per day.
Of those signals, 71.32% were profitable.
The average winning signal ran +11.2%, the average loser -2.5%.
Heimdall is not optimized for a high win rate, to be clear. This was a very good week. However while Heimdall isn’t always right, when it is, it’s right to a far larger degree than when it’s wrong.
What We Traded
Wednesday after close Heimdall said FPS was going to keep running off the guidance raise. Six people, +117% combined.
Same afternoon it called BTDR off the AI datacenter lease and three people bought calls. Three for three, 67% was the top, 110% combined.
DSP got called Friday premarket after the offering got pulled overnight and paid 32%.
And it’s not just calls. LCID got called to fade Thursday and the puts paid 76%. TSLA got called down into the delivery report and the puts paid 81%. VALE was 50% combined, HOOD 15%, S 10%.
It’s calling squeezes so well that I’m starting to learn new information about what they look like for the first time in years.
What Was Missed
Wednesday afternoon Heimdall called LGVN after the trial failed. It was down 64% by Friday. Thursday morning it called MKDW and that was down 61% the same day. Friday premarket it called DCX and that was down 58% by the close. DAIC was called Thursday and dropped 27%. BDRX 24%. TPST 23%. SION 23%. TCRT 22%. OPTT 21%. HCWC 20%.
Twelve more plays I found moved 20% or more in the direction Heimdall called them, options hundreds of % up and nobody had a position in them. Once everyone’s reading Bifrost the way we used to read Catalysts, this is what changes next. I said this would retire people, I’m still saying that.
Like I said last week, it’s as much about the losses as the plays. It also told us nine times that a play was dead this week, CIEN, INTC, a few others, called and then invalidated when the reason stopped holding, so we could stop holding.
Monday
Trading-floor is full of green again and this time it doesn’t fade, because it doesn’t depend on any one of us burning ourselves out. Three of us traded a week ago the way it used to take a full server to do. This week it was twelve, still a fraction of what it used to take and on average people were greener and safer than when it was a full sever.
And just a reminder, Heimdall is limited to reduce cost right now. This is only the second full week trading with it. Can you imagine as we start to learn it? As activity picks up with 100+ people? That’ll be sooner than you think. Can’t wait for Monday to continue this with you all, enjoy your weekends.
4 Likes
One emerging strategy that is somewhat prevalent is Heimdall catching pennystocks setting up for dilution very early in the morning before the dilution actually hits whether it’s by identifying a scammy precursor like a bullshit guidance raise, having taken a caymen exemption or disclosing funding that is actually just a shitcoin. DCX, UFG were examples of this but there are several others from this week.
The secondary would be that there are a growing example set where Heimdall is able to capture that the market hasn’t fully priced something in like with LGVN. That signal provided several hours before a 64% drop.
While I think getting too focused on specific strategies defeats the purpose of Heimdall I also think that knowing its strengths will help us evaluate opportunities even outside these. So it’s discouraged to only play these types of things, but giving them attention when they surface and understanding what from them leads to success is probably good.
So that is to say that I’m thinking of ways to better highlight things like this in the UI and that reading signals in the early AM is likely important for some of us to get started on.
2 Likes
TF’s winrate today was 88.89% on signal based positioning. These types of days where we didn’t see much were exactly the ones that would lead to the biggest red in our history. Strong start to the week off the heels of the 82.4% overall winrate logged last week.
1 Like
End of Week Wrap-Up
Week three was a wild one. Started off with the highest single-day winrate recorded and then mid-week hit the lowest. Remember that risk asymmetry I touted last week? Remember that “there are always new plays” I keep harping on about. Without further ado:
Some Stats
Positions
This week 16 people used the positions system to log 76 trades based on Heimdall’s signals.
13 of those people ended the week green.
Together they made over +677% on signal-linked trades.
They had an average single trade gain of 29.3%.
5 of those people are on a Free Trial.
They were profitable across 29 total signals, so 81.3% of signals they traded.
This makes for a two week streak at greater than 80%.
Of 18 logged trades without Bifrost coverage, only 50% of those trades were profitable.
Signals (IV Mispricing Removed)
Heimdall put out 184 signals across 165 tickers this week for an average of 37 per trading day.
The average winner is +6.7%, the average loser -2.9%.
Like I said last week, Heimdall is not optimized for a high win rate, the true winrate of the model this week was 58.5%. There was a pretty substantial bug affecting Tuesday and Wednesday but I won’t bother adjusting for it because it’s not the point. The point is that even still, TF maintained it’s reliable winrate across the turbulence because of the collaboration and most people ended the week green because of the asymmetry in risk still holding this week.
What We Traded
Tuesday, two minutes before the bell, Heimdall flagged GRRR. 20% of the float short, a fresh NVIDIA B300 deployment contract out of Taiwan, and nobody positioned for it yet. Six of us bought calls. Five of six green, +60% at the top, +93% combined.
ZS: restructuring and layoffs announced after close, right after a +20% five-day run. Called it down. Puts 2 for 2, +82% top.
CXW: CEO resigned for health reasons Friday premarket and the stock dipped 3%. Called the dip a buy. Calls 2 for 2 the same day, +65% top.
UPST: called the partnership pops to fade again. Puts +65%.
CTAS: guidance raise, called the downtrend over. Calls +51%.
MCD: traffic warning. Three of us, 4 for 4 on puts.
WOR +42%. FSLY +36%. BB +27% into earnings. SBUX +21%.
What Was Missed
Thursday it called GCTK’s 43% spike to unwind back to the $2.04 offering price. It closed today down 37%. Friday morning it called JAGX’s royalty pop to bleed into its brand-new ATM, down 26% the same day. BENF fade, down 26%. MX called long off Navitas buying in at $3.42, up 20%. LOBO fade after a 69% spike, down 20%.
Twelve calls nobody touched are still 10% or more in Heimdall’s direction at today’s close. Once everyone’s reading Bifrost the way we used to read Catalysts, this is what changes next. I said this would retire people. I’m still saying that.
When It Changed Its Mind
Like I said last week, it’s as much about the losses as the plays. Eight times this week Heimdall came back and said its own call was dead. FRMI was bullish Tuesday on a $6.5B data center lease; Thursday it sold off 13.5% to 52-week lows and Heimdall dropped the call, because the market was pricing in the dilution needed to fund it. RKLB, ONDS, SKHY, ALB, ACN, BB: called, then dropped when the reason stopped holding, so we could stop holding.
We Learn What Heimdall Is Good At
Heimdall’s worst calls this week were fades on small caps that squeezed instead. AVAT and GRML ran 60–70% against the call.
By working together we identify things the model isn’t good at and can size appropriately when we thing something in that category surfaces that we think is good. We learn to treat it like the gamble that it is.
Monday
Trading-floor is full of green again and this time it doesn’t fade, because it doesn’t depend on any one of us burning ourselves out. Three of us traded two weeks ago. This week eighteen of us opened positions off Heimdall’s signals, still a fraction of the old server, and 13 of the 16 who closed trades finished green.
This week I announced a completely revised role system, next week I have even larger announcements about features in Heimdall that I’m unlocking. I told you that this wasn’t as good as it gets and you’re going to see just how true that is as I start to unchain more of the system and Heimdall starts spending more of it’s time with it’s eyes open.
So pumped for Monday
2 Likes
Roles went live today, it may be a day or two between people start getting the quicker-earned ones. The ladder will take ~two weeks before people start climbing based on current figures. A note is that the entire role system is definitely in the testing stage, there will be bugs and tweaks needed.
Most of today is going to Heimdall work so I’m going to be quieter than I’ve been in the past week, since I’m working on something that if incorrectly configured will burn thousands per minute I’m sure everyone understands the caution.
Hope is to have clusters fully released alongside an unchained Heimdall by tomorrow alongside some enhancements that resulted from a significant amount of backtesting I conducted this weekend.
3 Likes
So while I have a brief sec I did want to share something regarding earnings from Heimdall’s backtesting this weekend.
- Recent stock performance has no bearing on earnings direction. As in there doesn’t seem to be any correlation between stock thats run a bunch with a mild beat and whether or not it continues to run as a result.
- News before earnings is not indicative of whether or not the earnings will be bad.
- Guidance effects are extremely one-sided. Cuts are punished, raises have some effect based on size but it’s to a far lesser magnitude than the profitability and revenue figures.
2 Likes
Heimdall signal quality should be increasing as of 9:30AM or so. Signals before that from today should be vetted thoroughly, also vet signals after that but you know what I mean.
2 Likes
Didn’t get a wrap up this week will get one up next week.
Tons of improvements this week and I’m working continuously this weekend. Today I deployed a ton of changes to the frontend optimizing speed, load times, form submits and making the majority of data streaming or live updating so the browser should no longer need to be refreshed for updates.
State of the model dropping tomorrow as well.
3 Likes
Lots of changes this week so I haven’t kept up much on announcements. Below are several UI changes for Bifrost. Next week we’ll be seeing the rollout of customizable notifications, including the first test of the “positions protection” alerts from the universe when things change on a position you have logged in Bifrost. Other notifications will alert you based on what the community is looking at and researching as well as the release of the first features that allow you to leverage Heimdall for searches and other information from Discord.
Heimdall
Heimdall is now the default page when Bifrost loads, Workspace can still be accessed via the navigation menu.
Driver Signals
Driver level signals are now visible in the Heimdall arc list
Drivers
Drivers now have shorter titles and are broken out into categories to make them easier to browse and understand. There is still more work to be done on this but I think we’re heading in the right direction.
Workspace
You can now save your custom boards and share them with the community. Workspace overall is getting some updates because it lags behind Heimdall in development so I’ll announce more on that when those changes land.
2 Likes
Month 1: Green for 4 Weeks Straight
Since Heimdall went live there has not been a single red week. Four weeks of green. AN ENTIRE MONTH. That’s happened maybe twice in the entire history of this community. 29 people trading it, 20 OF THEM ARE GREEN. Most everyone that started using this is GREEN and the ones that aren’t are BARELY RED. In any retail community on earth that is unheard of. Of the 360 calls it made, 180 ran 5% or better. That’s retirement. That’s all you need. Over 50 of our trades went 100%+. Multiple went 200%. $47,000 in profit across TF this month on $725 of membership fees.
Read that again.
FORTY SEVEN THOUSAND DOLLARS off seven hundred and twenty-five.
Valhalla currently has a 76% win rate on trades taken on signals. The average winner runs 41%+ and the average loser only goes -15%. Trades taken without a Heimdall signal lose more, about -33% on average, and win less, about 59%. Same people. Same Server. The only difference is Heimdall calling the plays so we’re consistently trading properly.
Heimdall is taking people who were never our top traders and is enabling them to put up numbers our upper crust used to pray for, just by reading and collaborating. Heimdall will retire people. People in this server are going to retire off this system if they stick with it. That’s a promise, write it down. If you can get that on Polymarket or Kalshi, list it and buy fucking calls, or whatever it is you actually buy on those stupid ass sites.
If this were 2022 you would be losing your fucking minds right now. Calling your friends. Calling your family and telling them to never join Valhalla and that if you die, please don’t open your Discord. Then you’d be on trading floor yapping because you had to know what happens next and didn’t want to miss something. Instead I’m watching people come out for VC night, nod, and go back to sleep like this isn’t a moment.
This IS a moment. This is AMC in May 2021, this is IRNT whenever I happened to call that shit, this is RSX with something that tells you to cut the puts. This is the exact moment most of you stayed in Valhalla for. The moment we got back to that. The moment you get the chance to do it again, to trade like you should’ve traded and read like you should’ve read.
Every generational wealth call you’d hoped Valhalla would produce gets dropped AT LEAST once a week by Heimdall. The Valhalla you sometimes think about and have nostalgia for isn’t gone. It’s right here running at full speed with a select group of people starting to make those life-changing gains you told yourself you’d never miss out on, and YOU ARE MISSING IT.
I’ve done my part. I need the community to do its part. The thing missing from Valhalla right now isn’t plays, features or the system. Those are built, they’re running and winning and they’re getting more insane by the day. What’s missing is you. Four weeks of green and trading floor goes quiet at 4PM like it’s a Tuesday in a bear market. There are people in this server that have barely read a single signal or even opened Bifrost. You’re watching this like it’s someone else’s server. It’s not. Valhalla isn’t me. Valhalla is US. If you want Valhalla back, the only part that’s missing is you and that fire. Right now, if the 150+ people I give a shit about that are getting this message start actually showing up every day, Valhalla of 2021/22 is back. Full stop. That’s it, that’s all it takes. I’ve built it, you just need to cum.
That’s the only difference between now and a trading floor full of people yelling about the same candle and think tanking the next winner at the same time. Every piece of this was built so that room could exist again, in a way that’s sustainable and profitable. It’s built and I knocked it out of the park. It’s just waiting on you. So get excited. Get fucking hungry. Turn those notifications back on, get on TF and participate, catch up, let’s have fun, and then the next day let’s get in Bifrost and find the next win, because there is one every day and it’s all wasted if we’re not working together to find it.
That goes for everyone. I’m one man. Together we are Valhalla, and it’s time to come home.
Monday at 8AM I’ll be in VC for the Morning Show: Discord
Have a great weekend. A lot of you have DMs from me through the Bifrost bot. Some of them die tonight at midnight. The rest die Wednesday.
1 Like
Another Heimdall backtest result that is of interest. @Navi asked why there weren’t any signals related to the current Hurricane. The reason is that most of the news that references it, Heimdall doesn’t deem meaningful (working as intended). But out of curiosity I ran some backtests:
At first this looked promising, but when I had Heimdall time travel (spins up a throwaway universe in a specific time and replays news and data leading up to a specific point) it’s just noise.
The issue is that not every Hurricane or storm actually makes landfall combined with the fact that even when you do nail that a storm will hit, it being bullish or bearish isn’t even guaranteed. For example for insurers they’ll sometimes fall obviously due to losses but they also sometimes climb due to forecasted premium increases. Then for refiners they’ll go down for lost throughput, but up likely because of higher margins at what facilities they have left.
For example refiners were +15% after Katrina and -5% after Laura.
Some of it is likely the weather data I’d imagine since institutions would be using non publicly available data but even with it, the edge window would still be very small I think and still probably amount to a coin flip.