Technology ยท Information Technology Services ยท Completed evidence
VNET relative to GDS
VNET Group Inc DRC is the relatively depressed leg. GDS Holdings Ltd is the relatively extended leg.
Specific pair strategy
What is the play in VNET / GDS?
This is a relative trade. The result depends on which security performs better, not on whether the whole market rises or falls.
Watchlist pair - not triggered
Possible future structure: long VNET / short GDS
The proposed pair would look for VNET to outperform GDS, but the current evidence does not yet justify treating the gap as a convergence trade.
This is the relatively depressed leg. The case needs it to recover or at least hold up better than the other leg.
This is the relatively extended leg. The case needs it to lag or weaken relative to the long leg.
A favourable relative move means VNET rises more than GDS, falls less than GDS, or rises while GDS falls. If GDS outperforms VNET, the pair moves against the case.
Wait for a completed daily turn toward normal and fresh evidence that VNET is improving relative to GDS. The size of the gap alone is not the trigger.
If the daily gap keeps widening, leave the pair alone rather than averaging into the separation.
Do not use the pair if the completed gap keeps widening, VNET continues to weaken relative to GDS, or the relationship moves into a break state.
Treat this as a watch condition until the spread turns and at least one leg develops confirming evidence.
Evidence now
- The spread itself has not yet gained a separate evidence catalyst.
What would invalidate it?
A large z-score alone is not a reason to expect mean reversion; the separation may reflect new information.
Possible setup paths
How the relative-value idea could be expressed
These are research structures, not recommendations or position-size instructions. The current course above determines whether any structure is ready to study now.
Long VNET / short GDS
Treat both legs as one relative position. The setup looks for VNET to outperform GDS; VNET can rise faster, fall less, or rise while GDS falls. The pair loses ground when GDS outperforms VNET.
When it becomes relevantThis is a structure to study only after the current course changes from โWatch, do not assume convergenceโ to confirmed convergence. Wait for the daily gap to narrow and for VNET to improve or GDS to lose support.Principal riskRequires a live review of both legs, borrow availability and cost, event dates, liquidity and the research normalisation.Study VNET alone if its own evidence confirms
This simpler expression focuses on recovery in VNET. It avoids short-borrow mechanics, but it no longer isolates relative value: the result is exposed to the broad market, the sector and VNET-specific news.
When it becomes relevantA stronger case needs improving VNET price structure and supporting evidence, not just a wide relationship gap.Principal riskCarries full single-security and broad-market direction risk.Study GDS alone if its support continues to weaken
This expression focuses on weakness in GDS. It may fit a leader-rollover case, but it also abandons the relative hedge and adds direct market direction, squeeze and company-event risk.
When it becomes relevantWeakening in GDS must be visible in its own completed evidence; relative extension alone is not enough.Principal riskShort selling can involve unlimited loss, borrow fees, recalls and forced close-outs.Any practical expression needs an independent review of suitability, portfolio risk, liquidity, current news and earnings, borrow availability and cost, transaction costs and tax.
Relationship gap ยท technical name: standardised residual
How far apart are they compared with normal?
Zero is the pair's historical centre. The outer guides show increasingly unusual departures from it.
This is the relationship gap after allowing for the pair's usual co-movement. Zero is its historical centre; the distance is measured in standard deviations, so different pairs can be compared on the same scale.
The latest gap is 0.94 standard deviations from normal. VNET is the relatively depressed leg and GDS is the relatively extended leg. That is unusual separation, not proof that either security is cheap or expensive.
A move toward zero means convergence. A move farther from zero means the separation is extending and raises the risk that the old relationship is changing.
Common-base price paths
Which security created the gap?
Both paths begin at 100 so their relative movement is visible without mixing different share prices.
Both securities are reset to 100 at the first date. This removes the distraction of different share prices and shows how their completed weekly price paths diverged. It is not a valuation chart.
Since the common starting point, GDS is +167.3% and VNET is +85.6%. The chart shows which price path contributed to the separation; the fitted relationship measure is more precise than the raw visual distance.
For convergence, watch for VNET to catch up, GDS to lose ground, or both. The relationship can close even if both securities rise or both fall.
GDSVNET
Completed daily timing
Is the gap closing yet?
The daily relationship adds timing context inside the slower weekly case. It does not replace the weekly validity checks.
This applies the same relationship to completed daily bars. It is the timing layer inside the weekly case, not a separate trading signal.
Gap extending: The completed daily gap is still widening. The weekly separation may be interesting, but timing has not confirmed a turn.
The first useful change would be for the daily line to stop extending and begin moving back toward zero.
VNET ยท VNET Group Inc DRC
14/100 evidence score- Market Dynamics
- -1.01 ยท change -0.141
- Trend
- Negative
- Stage
- Stage 4 ยท Declining
- Smart Money
- Neutral
- News sentiment
- Negative
- Options
- Bullish
GDS ยท GDS Holdings Ltd
21/100 evidence score- Market Dynamics
- -0.562 ยท change 0.543
- Trend
- Negative
- Stage
- Stage 4 ยท Declining
- Smart Money
- Neutral
- Short interest
- Crowded
- Options
- Bullish
Relationship diagnostics
Why this pair is on the map
How closely completed weekly returns moved together across the measurement window.
How consistent the fitted relationship remained across earlier and later parts of the sample.
An observed mean-reversion clock, not a deadline or a forecast.
How often the residual crossed its historical centre; repeated crossings matter more than a single fit.
The latest residual's location inside its own completed history.
Stored price, volume and size context only. It does not include a live short locate or cost.
State history
What changed and when
Daily timing changed
Completed daily timing changed to gap extending through 2026-08-21.
Research state changed
Changed from Balanced relationship to Unconfirmed stretch.
Daily timing changed
Completed daily timing changed to gap extending through 2026-08-18.
Daily timing changed
Completed daily timing changed to gap narrowing through 2026-08-17.
Relationship discovered
Published as Balanced relationship after the completed 2026-08-14 observation.
Same-industry map