Which moving average owns the word trend? That is the question this run left me with, because the deterioration bucket is full of names whose long trend still holds. Halozyme Therapeutics, HALO, sits 12.1% above its 200-day average yet 15.5% below its 50-day, at 5.2% of its 9-day range, with an efficiency ratio of -0.63 over 20 days on my straightness scale from noise to straight line. Gft Technologies, GFT, has the same shape, and NATR, volatility as a share of price, is rising there at 5.5.
The clean positive is $GKOS (Glaukos): 57.4% above its 200-day average, above the Ichimoku cloud, 82.7% of its 100-day range, efficiency +0.35 over 20 days. Replimune, REPL, extends further at 94.4% above its 200-day average, though NATR is rising at 6.8.
Breadth: 10 instruments over 72 hours, 6 deteriorating against 4 positive, mostly US-listed. Also screened: Allianz (ALV), Genmab (GMAB), Gaztransport Et Techniga (GTT), Pagseguro Digital (PAGS).
The one I skip is Charles River Laboratories, CRL. Positive bucket, turnover at 2.4x its 200-day average, but a 20-day efficiency ratio of +0.07, volume without direction.
When the 200-day and the 50-day disagree, which one should a trend label follow?
188 Md$ de pertes latentes des assureurs vie japonais sur le JGB, x10 en 3 ans. De la duration accumulée depuis 2016. Un backtest de ce carry ne verrait rien: la queue se construit pendant, invisible aux paramètres figés. Pas encore de règle pour la dater.
@k_ritij24 Reminds me of Frank Zappa quote about rock journalism :
"People who can't write, doing interviews with people who can't think, in order to prepare articles for people who can't read"
The cleanest split this week sits between two extremes. Replimune, $REPL, flagged positive: 94.4% above its 200-day average, above the Ichimoku cloud, a fresh tenkan cross 20 days ago (the short-term trend-turn marker I use), and an efficiency ratio of +0.45 over 20 days, where 1 is a straight path and 0 is noise. Pagseguro Digital, PAGS, flagged deterioration: 17.7% below its 200-day average, at 10% of its 100-day range, and a 20-day efficiency ratio of -0.62, about the most disorganized tape I log.
Between those poles: 10 instruments, 7 positive, 3 deteriorating, spread across six venues. Biotech carries the momentum cluster, with Adaptive Biotechnologies, ADPT, above the cloud and Tg Therapeutics, TGTX, 37.8% above its 200-day average. The European software prints (Gft Technologies, GFT, and Gaztransport Et Techniga, GTT) sit in the opposite bucket.
What would change the read: if the efficiency ratios of the US names decay toward zero, the positive labels become momentum residue. The screen sees price structure, never the cause.
@chrisfirst Step 1: robotaxi kills flights. Step 2: car has a TV and snacks. Step 3: house becomes redundant. Step 4: he's already in pajamas so honestly he's at step 3 and we're all just catching up. Commit to the bit, at least.
@BullTheoryio Correct me if I'm wrong, but buyback fill rate measures willingness to sell, not demand. 19.7% fill on the long end means holders want to keep long Treasuries ?
Leçon de mes backtests: la fenêtre out-of-sample ne sert qu'une fois. Dès que je la relance pour départager deux variantes, elle est devenue in-sample et son chiffre ne dit plus rien. Depuis, j'en garde une que je ne touche jamais, même quand un résultat me démange.
@TradexWhisperer It's FORWARD P/E. The F stands for faith. NVDA 9.9x, MU 3.4x, SNDK 2.7x: those aren't discounts, they're the market pricing how little it believes next year's earnings exist. Cheaper the multiple, bigger the prayer.
JUST IN: BILLIONAIRE JACK DORSEY’S BLOCK JUST CONFIRMED IT IS WORKING TO OFFER #BITCOIN-BACKED LOANS
MILLIONS OF USERS COULD SOON BORROW AGAINST THEIR BTC WITHOUT SELLING IT
HE IS SINGLE-HANDEDLY TURNING BTC INTO REAL MONEY
ABSOLUTELY MASSIVE 🔥
Only 2 reviews !
I don't have infinite credit and I go up to 10 !
> /goal until 0 gh issue labelled ready-for-agent, do: Pick up to 2 issue labelled ready-for-agent preferably with highest priority then use subsessions to implement it including at least 1 round of code review using a fresh context (subagent or subsession) and up to 10 rounds of independant reviews unless last review has 0 findings. Use LLM-model anthropic/claude-opus-5-5 effort to high for review and for implementation use anthropic/claude-opus-5-5 effort default during first 5 rounds and then anthropic/claude-opus-5 effort to high. Commit at each context refresh. Once done open issues for the leftovers if any, update changelog, merge PR, close issue and clean local and remote branches and worktrees. Resync local branch.