Indices

[Horizon: H5 Close]

BUY

TA-90 TA90

Forecasting and modeling

Model input: TASE / Yahoo Finance
Chart feed: TradingView / exchange feed
Rates: N/A | Macro sentiment: GDELT Project

Latest reference level 3,923.48

Active-horizon target (H5 forecast) 4,043.14 +3.05%

Raw interval (bands) S1 3,769.25
R1 4,077.71

Volatility (σ) 28.4% Previous completed assessment (H5): 4,293.57

Model quality (out-of-sample) 84.9% Out-of-sample sample: 200 sessions

Why this asset matters

TA-90 tracks the next layer of Israeli equities below the flagship blue chips. It is useful when you want to see whether participation is spreading beyond the heaviest names or staying concentrated at the top of the market. In that sense, TA-90 often acts as a breadth and risk-appetite read for the local equity tape.

What usually moves this asset

TA-90 is usually influenced by local risk appetite, domestic liquidity, small and mid-cap breadth, sector rotation and the tone of the broader Israeli equity market. Because it is less concentrated than TA-35, it can reveal participation changes earlier.

How to read this page

Use the public signal on TA-90 as a currently displayed horizon close breadth tilt rather than a stand-alone strategic call. It becomes more informative when compared with TA-35 and TA-125.

Methodology and dashboard reading (Model Card)

Model architecture:
Tree-ensemble architecture (Random Forest / gradient-boosted logic) built on price structure, volatility regime, macro-relative features, cross-asset context and sentiment inputs. For non-FX assets, the model is not centered on rate differentials alone.

Horizons (Targets):
H5: Direct forecast for the 5-session close (H5).
T+30: Direct horizon forecasts for each trading day up to T+30, displayed as a path.
1M-1Y: asset_model_targets_1m_1y_nonfx

Validation:
Walk-forward testing on expanding history without future leakage. Retraining is performed inside each evaluation step.

Trading definition of the signal:

Entry: At spot price (snapshot before model calculation). Exit: At the 5-session close (H5).
Divergence: If the technical trend is falling but the signal is BUY, the model has detected a local oversold condition and is trading a rebound (mean reversion).
Execution risk: The signal is not investment advice and does not include sizing. The model does not account for broker spreads, slippage, or carry costs (swap/rollover for FX).
News risk: At extreme risk sentiment (GDELT), raw bands are not calibrated for outliers. The probability of fat tails increases.

1. Current market regime (Technical & Macro View)

Heuristic regime gauges

Strong growth

Neutral

Heuristic regime display for page context, not a price-return backfill.

Heuristic regime indicators

90.0 (Overbought)

Trend-bias proxyBullish

Macro / carry biasBelow SMA 50/200

Volatility (σ)28.4%

Display heuristics for current regime reading. Not a full classical technical-indicator calculation.

Volatility (σ)28.4%

Macro factors (Percentile Score)

Values are normalized from 0 to 100

Sentiment (GDELT NLP)

Tension indexN/A

StatusData unavailable

Source: GDELT Project API
Coverage: ~400 media / 24h
Updated: 25.10.2026 02:30 UTC
News-tone classification, not a factual confirmation of the event.

2. Deep quantitative analytics

Historical seasonality

This panel is temporarily hidden while historical monthly returns are being rebuilt from the same source-aligned series used by the model.

A real seasonality heatmap will return after monthly returns are recalculated from the same source-aligned history used by the model.

Medium-term continuation estimate

For indices, equities, bonds and commodities, the 1M-1Y block is shown as a continuation of the 30-trading-day ML anchor, not as an FX-style CIP fair-value model.

Inputs: 30-trading-day ML anchor + medium-term continuation profile
30D anchor: 4,698.44
Extension profile: Anchor continuation with volatility widening

Current level: 3,923.48
Medium-term scenario: 4,698.44

Mean-revertContinue

Gap versus 30D anchor: 0.00%

Model verdict: Balanced continuation

3. ML model assessment (Predictive Analytics)

Historical validation: H5 forecast vs reality

Bands are nominal (raw). Current empirical out-of-sample coverage: 82%. The model is still being calibrated.

For historical validation, this chart uses the same source as the model input: TASE / Yahoo Finance. This removes cross-source shift.

Return-space check: predicted vs realized H5 move

This view removes the false level-lag effect and shows the predicted change from the horizon start close.

ML model summary (tree ensemble)

Analyzing macroeconomic data, the ML model detects a confident bullish (uptrend) regime on the medium-term horizon. The news-sentiment layer is unavailable in this refresh and is excluded from the narrative interpretation. RSI (90.0) confirms the local structure.

H5 signal: BUY
Return to target: +3.05%

Attention (Risk calibration):
Raw interval. Empirical OOS coverage: 82%. Calibration is still required (fat tails).

Directional model accuracy estimate 61.5%

Empirical coverage 82%

4. Probabilities and trading signals

Close probability over a 30-trading-day horizon Mode: Normal (Calm)

Estimate of the probability that price closes above or below target levels exactly after 30 trading days (T+30 sessions). Calculated via Gaussian CDF N(μ, σ). Does not account for skewness or fat tails during geopolitical FX shocks.

Close ABOVE level (Upside)

Level R1 (4,077.71)97%

Level R2 (4,136.56)96%

Close BELOW level (Downside)

Level S1 (3,769.25)1%

Level S2 (3,710.40)1%

Trading interpretation of the signal:

Instrument: TA90
Entry: At current session close (04:30).
Exit: At active-horizon close (30.10).
Model: BUY (Expected return: +3.05%)

Signals are produced exclusively by the ML core and are not investment advice.

Attention: The daily forecast (left) is displayed as direct horizon estimates for each trading day up to T+30. The long-horizon block (right) is a structural scenario layer, so the trajectories can diverge.

Short-horizon ML expectations (daily)

Date	ML target	Raw interval (bands)	Signal
26.10.2026 (Mon)	3,947.12	[3,833.84 - 4,063.76]	BUY
27.10.2026 (Tue)	3,970.91	[3,810.70 - 4,137.86]	BUY
28.10.2026 (Wed)	3,994.84	[3,798.34 - 4,201.51]	BUY
29.10.2026 (Thu)	4,018.92	[3,791.53 - 4,259.94]	BUY
30.10.2026 (Fri)	4,043.14	[3,788.25 - 4,315.17]	BUY
02.11.2026 (Mon)	4,067.50	[3,787.47 - 4,368.24]	BUY
03.11.2026 (Tue)	4,092.02	[3,788.58 - 4,419.76]	BUY
04.11.2026 (Wed)	4,116.68	[3,791.19 - 4,470.11]	BUY
05.11.2026 (Thu)	4,141.48	[3,795.03 - 4,519.57]	BUY
06.11.2026 (Fri)	4,166.44	[3,799.90 - 4,568.35]	BUY
09.11.2026 (Mon)	4,191.55	[3,805.65 - 4,616.58]	BUY
10.11.2026 (Tue)	4,216.81	[3,812.18 - 4,664.39]	BUY
11.11.2026 (Wed)	4,242.22	[3,819.39 - 4,711.87]	BUY
12.11.2026 (Thu)	4,267.79	[3,827.21 - 4,759.09]	BUY
13.11.2026 (Fri)	4,293.51	[3,835.58 - 4,806.12]	BUY
16.11.2026 (Mon)	4,319.38	[3,844.44 - 4,853.00]	BUY
17.11.2026 (Tue)	4,345.41	[3,853.77 - 4,899.78]	BUY
18.11.2026 (Wed)	4,371.60	[3,863.52 - 4,946.49]	BUY
19.11.2026 (Thu)	4,397.95	[3,873.67 - 4,993.18]	BUY
20.11.2026 (Fri)	4,424.45	[3,884.19 - 5,039.86]	BUY
23.11.2026 (Mon)	4,451.11	[3,895.05 - 5,086.57]	BUY
24.11.2026 (Tue)	4,477.94	[3,906.23 - 5,133.32]	BUY
25.11.2026 (Wed)	4,504.93	[3,917.73 - 5,180.13]	BUY
26.11.2026 (Thu)	4,532.07	[3,929.52 - 5,227.03]	BUY
27.11.2026 (Fri)	4,559.39	[3,941.59 - 5,274.02]	BUY
30.11.2026 (Mon)	4,586.86	[3,953.92 - 5,321.12]	BUY
01.12.2026 (Tue)	4,614.51	[3,966.52 - 5,368.36]	BUY
02.12.2026 (Wed)	4,642.31	[3,979.35 - 5,415.72]	BUY
03.12.2026 (Thu)	4,670.29	[3,992.43 - 5,463.24]	BUY

Long-horizon structural expectations

Horizon end	ML target	Delta	Raw interval (bands)	Signal
25.01.2027 (1 quarter)	4,839.54	+23.35%	[3,836.61 - 6,024.29]	BUY
25.04.2027 (6 months)	5,059.17	+28.95%	[3,483.11 - 7,108.96]	BUY
25.10.2027 (1 year)	5,542.44	+41.26%	[3,134.16 - 9,102.36]	BUY

Long horizons are shown as a structural scenario layer built from spot, rates, volatility and macro drift. Read them separately from the short-horizon ML forecast.

5. Research sandbox (Illustrative sensitivity)

Stress-test calculator (What-If Analysis)

Illustrates linear sensitivity to shocks (based on heuristic weights). Under stress, the risk corridor widens (volatility).

Geopolitical shock (GDELT): 0

Crisis (-1)Peace (+1)

US Dollar Index shock (DXY): 0%

DXY down (-10%)DXY up (+10%)

Short-horizon ML forecast (up to 30 trading days)

Long-horizon structural scenario (up to 1 year)

6. Interactive chart (TradingView)

Attention: the base model calculation uses TASE / Yahoo Finance. The chart below shows the TradingView feed (TASE:TA90), so small quote differences are possible.

Related market pages

FAQ

Why track TA-90?

Because it helps show whether market participation is broadening beyond the very largest Israeli stocks.

How is TA-90 different from TA-35?

TA-90 covers the next layer of equities and therefore says more about breadth than concentrated blue-chip leadership.

How should I read an active-horizon signal here?

As a short-horizon tilt in Israeli mid-cap participation into the currently displayed horizon close.

Why can TA-90 diverge from TA-35?

Because market breadth and blue-chip leadership do not always move together.