That's some cool research you've got there. My question would be are Hammers and shooters treated as their own category since they are a candle shape and not a strat candle?
For example, an inside (1) bar can be a shooter or hammer. Same with a 2dg, or even a 2u momo hammer. How are these accounted for in the probabilities of candle transitions. If a shooter resolves as a hammer and/or 2dg, is it counted into both categories?
Great question @Andy - yes, hammers and shooters are treated as their own overlays, so a bar can be counted both by its Strat scenarios and by its shape. For example, a 1 bar that’s also a shooter contributes to the 1s row/column as well as to shooter stats. Same with a 2dG that forms a hammer or a 2u that shows momentum with a hammer structure. In short: yes, it’s counted in both categories.
A quick note on why: while hammers and shooters aren’t Strat scenarios in the strictest sense, Rob consistently highlighted them in his nightly vids. Because of that, I've modelled it to tracks them alongside Strat scenarios. Longer-term, there’s a dedicated study on the backlog to test whether hammer/shooter tags add alpha on top of the base transition matrix, especially momentum hammers.
Speaking of momo hammers, I've a project in flight to review past 100%+ runners over the last 10 years and reverse engineering scans that would have surfaced them. This I've found is relatively easy. The harder problem is what's the likelihood that I would have trader them.
The bit I'm currently grappling with is to rewind the time capsule to that same morning to see what other names are in the basket and the likelihood that I would have picked it. What attributes could have best predicted followthrough? The goal is to fold those features into a lightweight scoring engine. From what I’ve seen, momo hammers tend to show up often in these leadership names. So I’m leaning towards including them in the scoring criteria.
Great discussion. I've sent you a friend request on the strat Discord btw (BlueMaroon). I'd love to discuss more if you're amenable. Would you ever do a follow-up study looking at the H/D timeframes candle (2dG, 3U, 2u, etc) probabilities of going 2U, when utilizing FTFC? I'm curious to know whether there are significant increases in % if lets say the D, W, M, and/or Q are also green. Perhaps even if they the higher TFs are "in-force" as well?
I'm sure you have no shortage of ideas to look into, and don't mean to throw more on your plate. Just interesting areas I'm curious about to see if there is truly an edge in these factors.
Thanks for the feedback, Andy. Snowman asked for something similar too. I’ll finish up the bearish pivot combos first and then circle back to this in October. Personally, I find the hourly too noisy, but daily actionable signals when the month is in force is definitely on the cards.
Keep the ideas coming. :)
On FTFC, I’m finding the rate of change approach unreliable. Instead of comparing the open a month ago vs. close today, I’m planning to test anchoring VWAP from a month ago against today’s VWAP or close measure TFC.
Once I wrap up those two pieces, I’ll also try to model AS on the daily when the month is in force.
That's some cool research you've got there. My question would be are Hammers and shooters treated as their own category since they are a candle shape and not a strat candle?
For example, an inside (1) bar can be a shooter or hammer. Same with a 2dg, or even a 2u momo hammer. How are these accounted for in the probabilities of candle transitions. If a shooter resolves as a hammer and/or 2dg, is it counted into both categories?
Great question @Andy - yes, hammers and shooters are treated as their own overlays, so a bar can be counted both by its Strat scenarios and by its shape. For example, a 1 bar that’s also a shooter contributes to the 1s row/column as well as to shooter stats. Same with a 2dG that forms a hammer or a 2u that shows momentum with a hammer structure. In short: yes, it’s counted in both categories.
A quick note on why: while hammers and shooters aren’t Strat scenarios in the strictest sense, Rob consistently highlighted them in his nightly vids. Because of that, I've modelled it to tracks them alongside Strat scenarios. Longer-term, there’s a dedicated study on the backlog to test whether hammer/shooter tags add alpha on top of the base transition matrix, especially momentum hammers.
Speaking of momo hammers, I've a project in flight to review past 100%+ runners over the last 10 years and reverse engineering scans that would have surfaced them. This I've found is relatively easy. The harder problem is what's the likelihood that I would have trader them.
The bit I'm currently grappling with is to rewind the time capsule to that same morning to see what other names are in the basket and the likelihood that I would have picked it. What attributes could have best predicted followthrough? The goal is to fold those features into a lightweight scoring engine. From what I’ve seen, momo hammers tend to show up often in these leadership names. So I’m leaning towards including them in the scoring criteria.
Great discussion. I've sent you a friend request on the strat Discord btw (BlueMaroon). I'd love to discuss more if you're amenable. Would you ever do a follow-up study looking at the H/D timeframes candle (2dG, 3U, 2u, etc) probabilities of going 2U, when utilizing FTFC? I'm curious to know whether there are significant increases in % if lets say the D, W, M, and/or Q are also green. Perhaps even if they the higher TFs are "in-force" as well?
I'm sure you have no shortage of ideas to look into, and don't mean to throw more on your plate. Just interesting areas I'm curious about to see if there is truly an edge in these factors.
Thanks for the feedback, Andy. Snowman asked for something similar too. I’ll finish up the bearish pivot combos first and then circle back to this in October. Personally, I find the hourly too noisy, but daily actionable signals when the month is in force is definitely on the cards.
Keep the ideas coming. :)
On FTFC, I’m finding the rate of change approach unreliable. Instead of comparing the open a month ago vs. close today, I’m planning to test anchoring VWAP from a month ago against today’s VWAP or close measure TFC.
Once I wrap up those two pieces, I’ll also try to model AS on the daily when the month is in force.