Daily brief

Wednesday, November 18, 2026

What do you need to know today? This page is the 90-second version: the headlines that matter, what the model is doing and why, and the dates coming up that can move stocks.

General commentary from public data, not personalized financial advice.

the brief
2026.11.18 // 07:10 utc
BMY

BMY has a late-stage trial wrapping up tomorrow

Sjögren's Syndrome
BMYtomorrow
MRKon Friday
GSKon Monday
The dates are public. The results are not.
NVDA

All the questions Nvidia should answer on Tuesday — but probably won’t

NVDA
TechCrunch
OKLO

Sell OKLO at the next close

model, since March+1%
market+15%
Day 5 is sell day, win or lose.
The model's calls
What is the model holding right now, and why? Every position with its printed reason.
Since Mar 26, $10,000 became $9,619. The same money left in a market fund: $11,412. 47 of every 100 closed trades made money (199 trades closed so far).
PATHday 0 of 5+0.0% so far
Bought Nov 17 at $12.39, now $12.39. Sells automatically after 5 trading days.
MSFTday 1 of 5-2.3% so far
Bought Nov 16 at $517.99, now $506.24. Sells automatically after 5 trading days.
MSFTday 1 of 5-2.3% so far
Bought Nov 16 at $517.99, now $506.24. Sells automatically after 5 trading days.
PATHday 1 of 5-5.7% so far
Bought Nov 16 at $13.14, now $12.39. Sells automatically after 5 trading days.
MSFTday 2 of 5-3.4% so far
Bought Nov 13 at $523.88, now $506.24. Sells automatically after 5 trading days.
MDBday 3 of 5-12.1% so far
Bought Nov 12 at $339.78, now $298.63. Sells automatically after 5 trading days.
MSFTday 4 of 5-2.9% so far
Bought Nov 11 at $521.13, now $506.24. Sells automatically after 5 trading days.
Reason: that day's top-scoring story. UGG: Unified Generative Geometry Model
How it decides: each trading day the model puts $2,000 on the day's top Attention story and always sells after 5 trading days. The formula decides, not a person.
Follow these in My Picks, one table →
Every trade it ever made →
Dates that can move stocks26
Which scheduled events coming up could swing a stock? Public dates, unknown outcomes.
Scheduled events with public dates and unknown outcomes. Informational only, not investment advice.
Did the calls make money?
The running score for every call this site flagged in the last year, losers included.
47 of 100
calls beat the market within a week
Out of 5,089 calls measured, 47 of every 100 did better than the market fund over the 5 trading days after the call.
34 of 100
calls jumped at least 2% within a week
34 of every 100 flagged names rose at least 2% in the 5 trading days after we flagged them.
-1.0%
vs the market after a month
The average call finished behind the market fund 20 trading days later, across 4,578 calls measured.
Best call in the window: SPCE up 10.9% in the week after (May 1).
Worst call in the window: SPCE down 45.2% in the week after (Jun 1).
The full score sheet →
In the news30
What actually happened out in the world today? Real stories from the wire, tied to tickers where we can.
California state AG sues OpenAI over rogue agent incidents
TechCrunch · Nov 16 · AI & softwareMSFT
The full wire →
What fed the scoring5
Which research stories scored highest today? Attention is our 0-to-10 measure of how much signal a story pulled across sources. The model trades the top one; the rest are context, not calls.
Attention
21.0
MSFT$506.24 (-2.27%) GOOGL$280.20 (-1.24%) META$618.87 (-1.20%) NVDA$155.46 (-0.44%) novelty spike·multi-ticker·large-cap exposure·benchmark lead·efficiency/cost
Why it matters, Rising momentum suggests near-term attention and follow-on activity.
Why it matters, Impacts multiple players/supply chain; effects may propagate.
Attention
19.7
MSFT$506.24 (-2.27%) GOOGL$280.20 (-1.24%) META$618.87 (-1.20%) ARM$183.39 (+4.76%) NVDA$155.46 (-0.44%) OKLO$22.37 (-2.87%) NNE$10.49 (-2.33%) multi-ticker·large-cap exposure·efficiency/cost·real-time/edge·safety/alignment
Why it matters, Impacts multiple players/supply chain; effects may propagate.
Why it matters, Large-cap ties can amplify market impact and adoption.
Attention
19.6
MSFT$506.24 (-2.27%) GOOGL$280.20 (-1.24%) META$618.87 (-1.20%) NVDA$155.46 (-0.44%) MDB$298.63 (-5.77%) ARM$183.39 (+4.76%) PATH$12.39 (-5.71%) BBAI$1.80 (-1.10%) multi-ticker·large-cap exposure·benchmark lead·efficiency/cost·real-time/edge·data/training
Why it matters, Impacts multiple players/supply chain; effects may propagate.
Why it matters, Large-cap ties can amplify market impact and adoption.
Attention
16.1
MSFT$506.24 (-2.27%) GOOGL$280.20 (-1.24%) META$618.87 (-1.20%) RKLB$55.05 (-4.04%) SPCE$1.30 (-16.67%) OKLO$22.37 (-2.87%) NNE$10.49 (-2.33%) novelty spike·multi-ticker·large-cap exposure·efficiency/cost·data/training
Why it matters, Rising momentum suggests near-term attention and follow-on activity.
Why it matters, Impacts multiple players/supply chain; effects may propagate.
Attention
15.2
MSFT$506.24 (-2.27%) GOOGL$280.20 (-1.24%) NVDA$155.46 (-0.44%) IONQ$24.84 (-6.09%) IBM$231.09 (-4.71%) multi-ticker·large-cap exposure·open-source release·benchmark lead·safety/alignment
Why it matters, Impacts multiple players/supply chain; effects may propagate.
Why it matters, Large-cap ties can amplify market impact and adoption.
The research radar
What new papers and patent filings crossed the scanner today? Raw inputs to the scoring above, not calls.
AX2026-11-17T18:56:09Z
Data-driven predictive control (DDPC) predicts system responses from recorded trajectories, but nonlinear dynamics can limit the accuracy of fixed linear predictors. This paper studies geometric robust DDPC for trajectory tracking of a fully actuated manipulator using model-based feedback linearisation. A mechanical model maps virtual acceleration commands into joint torques, while recorded data capture residual dynamics arising from imperfect model parameters. The controller minimises worst-case prediction error over an open Grassmannian ball centred at the fitted trajectory subspace. Simulations on a planar 2R manipulator under gravity compare the proposed method with double-integrator model predictive control (MPC) and direct regularised data-enabled predictive control (DeePC). Results show improved tracking over double-integrator MPC despite increased torque effort, while performance relative to DeePC depends on the reference trajectory.
AX2026-11-17T18:50:22Z
Human cognitive state (e.g., distraction, stress), gaze, and behavior interact with the driving environment in subtle ways that AI could use to assist drivers, but ground-truth labels for human cognitive state levels are costly to obtain at scale. We propose a multimodal masked-diffusion autoencoder, ID-MAE, that adds a self-supervised objective to impairment prediction: alongside predicting the impairment label as a masked token, it reconstructs masked gaze and trajectory. To probe the effects of different modalities and types of impairments on model inference, we introduce DIVERT, a driving-simulator dataset containing 103 hours of driving data (n = 109) with ten impairment inductions spanning cognitive, visual, and emotional states, recorded with gaze, vehicle state, and scene video. We demonstrate how ID-MAE affords significant improvements in impairment inference results on both the DIVERT dataset and existing datasets, and enables strong results in additional tasks, such as trajectory and gaze prediction. For location-matched pairs, our model achieves a paired AUC of 0.706 for impairment detection on DIVERT, compared to 0.590 without reconstruction loss, improving over existing baselines. We probe the design space in terms of input modalities, masking scheme, positional encoding, attention mechanism, and decoding. We also investigate the interaction between various impairments in training and inference, demonstrating that visually demanding distractions are separable from cognitive distractions. Code and data will be available upon acceptance.
AX2026-11-17T18:38:28Z
Trajectory prediction models for automated driving are usually trained on data with a fixed sampling frequency, while positions received via vehicle-to-everything (V2X) communication arrive at 1-10 Hz. We study whether a learned predictor that uses only the observation history keeps its advantage over constant velocity (CV) and constant acceleration (CA) models when the history frequency drops. On the Argoverse 2 motion forecasting dataset, the 5 s history of the focal agent was downsampled from 10 Hz to 5, 2, and 1 Hz, and the 6 s future was predicted without maps or neighboring agents. An LSTM-based recurrent neural network (RNN) trained for each frequency was compared with the same network trained at 10 Hz and applied to sparse histories after linear interpolation. Hypotheses and thresholds were fixed before training, the main RNNs were trained with ten seeds each, and results on 24,984 test scenarios have bootstrap confidence intervals over scenarios and seeds. The average displacement error of the frequency-specific RNN grew by 3.1% at 5 Hz, 10.1% at 2 Hz, and 20.3% at 1 Hz. Contrary to our hypothesis, its advantage over the better kinematic model did not vanish on straight motion and was 24-30% on both straight motion and turns at every frequency, while for pedestrians CV was more accurate. The 10 Hz model with interpolation was at most 1.3% less accurate than the frequency-specific model down to 2 Hz but 11.5% less accurate at 1 Hz. Each hypothesis was only partly supported.
AX2026-11-17T18:35:53Z
Goal recognition is the task of inferring an agent's intended goal from its observed behavior. Most existing work focuses on high-level symbolic recognition, abstracting away the continuous motion through which agents reach their goal. Methods that do reason over continuous trajectories rely on an explicit dynamics model that is hard to acquire and rarely transfers across settings. Neither approach explicitly learns how agents move toward each candidate goal, limiting accuracy on ambiguous trajectory prefixes where such knowledge is most needed. We present WoRD (Waypoint Recognition with Diffusion), which uses a learned trajectory prior to infer the agent's goal from partial observations. At each step, WoRD runs one warm-started denoising chain per candidate goal, steers each chain toward the observed trajectory through a closed-form log-likelihood gradient, and reweights the resulting samples by Bayes' rule into a calibrated posterior over goals. We evaluate WoRD on simulated robot settings, pedestrian trajectories, and vehicle trajectories using four metrics targeting distinct failure modes. Across all three settings, the same inference procedure preserves calibrated multimodal beliefs through ambiguous prefixes where filter- and classifier-based observers collapse to a single mode or grow overconfident.
AX2026-11-17T18:03:24Z
The agents in a heterogeneous team can cooperate to accomplish tasks that exceed their individual capabilities. The team trajectories for a cooperative task can be generated by a generative model, provided that the model has been trained on demonstrations of the agents cooperating. Yet, demonstrations of cooperation are costly and may not exist for tasks that the team has never performed. The conditions under which the agents can combine their capabilities, however, are physical properties of the team and are known before any demonstration is collected. We encode the conditions as constraints on the team trajectories, and we enforce the constraints while sampling from generative models trained on demonstrations of each agent alone. We prove that the proposed method can generate every team trajectory that satisfies the constraints, although the demonstrations contain no cooperation. Hence the constraints determine which team trajectories can be generated, and the demonstrations determine how likely each team trajectory is. Numerical results on two cooperative tasks show the effectiveness of the method, which generates cooperative team trajectories that satisfy the constraints and requires no retraining when agents join or leave the team. Finally, we implement the method on a multi-robot testbed.
PT2014-12-10
Method for teaching a robot movement (84 - 88 - 90 - 92) using a system comprising . a robot (36, 94), . a robot controller (34, 96) with at least an automatic mode and a teach mode, . a programmable logic controller (PLC) (32) which is connected (38) to the robot controller (34, 96), whereas the …
PT2019-05-07
The invention relates to a method for programming a robot, in particular a robot comprising a robotic arm, in which method a movement to be performed by the robot is set up preferably in a robot programme by means of a predefined motion template, the motion template is selected from a database …
PT2025-10-03
A robotic controller for controlling a robotic arm is disclosed, the robotic controller comprising: a first spatial shaping module configured to provide a shaped first-space target motion by convolving a first-space target motion with a pulse train, wherein the first-space target motion defines a …
PT2026-04-28
The numerical control system (1) includes a numerical control device (5) and a robot control device (6). The numerical control device (5) includes: a robot instruction generation unit (55) that generates robot instructions for each robot instruction block; a robot program start instruction unit ( …
PT2012-08-15
1. the mechanical arm controller of a tool motion control, logic control core is characterized in that, comprising: One mechanical arm is multivariant mechanical arm, and said multivariant mechanical arm uses a plurality of servo motor driven mechanical arms, to carry out movement locus control;