Making sense of 大混乱? Algos scour social マスコミ for 手がかり（を与える）s to crypto moves

• e-mail

By Tom Wilson and Simon Jessop

LONDON, July 17 (Reuters) - After months of 親族 静める in cryptocurrency markets, bitcoin 爆発するd 支援する into life in April with its はっきりした price jump in over a year - but few people could convincingly explain why.

The 20% leap 焦点（を合わせる）d 投資家s' attention on one of the 耐えるing mysteries of cryptocurrencies: what moves the price of an 現れるing 資産 in an opaque, 大部分は unregulated market?

For some, the answer lies on social マスコミ. Hedge 基金s and 資産 経営者／支配人s 捜し出すing an 辛勝する／優位 are training computers to 捨てる social マスコミ 場所／位置s for 誘発する／引き起こすs that could move the price of 数字表示式の 通貨s.

Their goal: （手先の）技術ing algorithms 有能な of 選ぶing out price "signals" from the background noise of 場所／位置s 範囲ing from Reddit and WeChat to Twitter and 電報電信.

Many 投資家s already use computer models to identify, and 貿易（する）, price differences across hundreds of cryptocurrency 貿易（する）ing 交流s.

But with 適切な時期s for arbitrage 狭くするing as the nascent 部門 develops, big players are ますます looking to build or buy more sophisticated robots to find market-moving signals online, によれば interviews with six hedge 基金s and 資産 経営者／支配人s and three ソフトウェア developers.

Yet while the use of algorithms, or algos, for parsing social マスコミ may be growing, some of those interviewed said major challenges and 危険s remain to their wider （軍隊などの）展開，配備, from cost to 複雑さ.

"It's an 武器 race for money 経営者／支配人s," said 貯蔵所 Ren, CEO of Elwood 資産 管理／経営, which specialises in 数字表示式の 資産s and is owned by Brevan Howard 創立者 Alan Howard.

"Very few players are able to 器具／実施する and 配達する it, but I believe it is 高度に profitable."

Such "感情 分析," as computer-driven reading of the social マスコミ mood is known, is used as a 道具 in 伝統的な markets like 公正，普通株主権s and foreign 交流 to 貿易（する） on 消費者 feelings に向かって a company or 資産.

It is also 早期に days for the technique in the crypto 部門, with scant 産業-wide data on 業績／成果 and many questions over its 有効性. 非，不，無 of the 会・原則s Reuters spoke to would give 詳細（に述べる）s of the 業績／成果 of their algorithms, 特記する／引用するing 商業の confidentiality.

ONLINE CACOPHONY

To be sure, 数字表示式の 通貨s do 株 some drivers with 伝統的な markets such as comments by policymakers. Bitcoin can be 極度の慎重さを要する to 発言／述べるs by regulators in particular: It fell はっきりと last week after the U.S. 連邦の Reserve 長，指導者 called for a 停止（させる） to Facebook's planned Libra cryptocurrency 事業／計画（する）.

But given cryptocurrencies have been entwined with the internet from their 夜明け a １０年間 ago, when the word was spread in 会議s and chatrooms, it would seem to make sense to search for price 誘発する／引き起こすs online.

Still, it's far from cheap or simple to design an algorithm that can find market-moving signals in the cacophonous world of social マスコミ, analysing 抱擁する numbers of 地位，任命するs in dozens of languages while 精査するing out unreliable （警察などへの）密告，告訴（状）.

Andrea Leccese, 大統領，／社長 of Bluesky 資本／首都, an 投資 会社／堅い in New York, said upfront costs for a robot 有能な of only reading Twitter in English were between $500,000 to $1 million, with most of the money spent on 技術d developers. That has deterred Bluesky from using the technique, he said.

One daunting challenge is the sheer number of social マスコミ channels. Beyond Twitter, 場所／位置s often used by cryptocurrency aficionados 含む 電報電信, a messaging app with public channels and Reddit, a messaging board.

In Asia, home to many 小売 仲買人s, apps like Line in Japan and Kakao in South Korea are popular.

Tens of tho usands of comments on cryptocurrencies are pumped out around the clock across both 国家の and international channels.

Reddit's main 会議, or subreddit, for bitcoin alone has 1.1 million members. Twitter also sees tens of thousands of 地位，任命するs について言及するing bitcoin every day, with between 14,000 and 32,000 daily for the last three months, によれば the BitInfoCharts website.

In an 試みる／企てる to 抽出する meaning from this mayhem, algorithms use いわゆる natural language 過程ing - identifying 重要な words and emotions that 示す changes in how social マスコミ 使用者s 見解（をとる） 確かな 数字表示式の 通貨s.

投資家s using algorithms say they can also identify patterns for （警察などへの）密告，告訴（状） that 伸び（る）s traction online.

"The （警察などへの）密告，告訴（状） propagates not 無作為に, but through a very 井戸／弁護士席-defined structure - it's like a tree," said Elwood's Ren, which has used 感情 分析 for nearly two years after developing its own ソフトウェア.

"It's very 類似の to modelling the spreading of a ウイルス."

FAKE NEWS FEARS

Other 投資家s 強調d the challenges in teaching machines to 位置／汚点／見つけ出す biased or 不確かの （警察などへの）密告，告訴（状）.

A Reuters 報告（する）／憶測 https://www.reuters.com/article/us-crypto-通貨s-promoters-specialre/special-報告（する）／憶測-little-known-to-many-投資家s-cryptocurrency-reviews-are-for-sale-idUSKCN1NW17S last November 設立する that many social マスコミ 使用者s take money for 肯定的な reviews of 数字表示式の coins.

BitSpread, a cryptocurrency 資産 経営者／支配人 based in London and Singapore, uses its own 資本／首都 to 貿易（する） using an algorithm it started developing about a year ago, its CEO Cedric Jeanson told Reuters.

It is a 比較して 辛うじて 的d ソフトウェア. Aggregating Twitter 料金d, it looks out for 地位，任命するs on the liquidation, or の近くにing, of positions at 交流s.

"It's a 事柄 of 集会 all the info, trying to understand who is 貿易（する）ing where, what 肉親，親類d of liquidation can appear," he said. "It's a 戦略 that makes sense."

However, he 定評のある the drawbacks.

"The sen timent itself, what we see on Twitter, can be really geared に向かって 偽の news. We are always very 用心深い about what we're reading in the news because, most of the time, we've seen that there's a bias."

Many algorithms use machine learning, where they are supposed to 改善する through experience and better understand how social マスコミ 地位，任命するs translate into market movements.

Developers often identify 重要な people with outsized 発言する／表明するs and large numbers of 信奉者s to 負わせる more ひどく in their algorithm, said Bijan Farsijani of Augmento, a Berlin-based startup that 開始する，打ち上げるd an algo for 感情 分析 last month.

He said a number of hedge 基金s had bought the ソフトウェア from his company since the 開始する，打ち上げる.

Bitcoin's wild ride - interactive graphic https://tmsnrt.rs/32kDEi2

CODERS IN DEMAND

Bitcoin, the biggest cryptocurrency and a bellwether for the 部門, has 殺到するd over 180% this year, 運動ing up the 利益／興味 of bigger 投資家s from 貿易（する）ing 会社／堅いs to hedge 基金s.

Bitcoin's most 最近の 決起大会／結集させる, last month, was seen by 分析家s as driven by 期待s for a wider 採択 of cryptocurrencies driven by Facebook's Libra.

That move was mirrored by a 殺到する in 利益／興味 online. Google searches for cryptocurrencies 攻撃する，衝突する their highest level in three months on June 18, when Facebook made the 告示.

It is, however, difficult to pinpoint the chicken and the egg: online chatter or price moves.

"There may be some value in 感情 分析 in crypto, but most of the time what people tweet may be a lagging 指示する人（物） of the price move," said Leccese of Bluesky 資本／首都.

"But there is 可能性のある," he 追加するd. "People will start looking at this more in the next five to 10 years because there will be 減らすing returns because of 増加するd 競争 in 伝統的な 戦略s."

While there is a 欠如（する） of data 特に for this technique, "quantitative" cryptocurrency 基金s - which use methods from arbitrage to 感情 分析 - 意味ありげに outperformed 基金s that make longer-称する，呼ぶ／期間／用語 bets in the first ４半期/4分の1 of this year, a PwC 報告（する）／憶測 shows.

Coders say they are in 増加するing 需要・要求する.

Taiwan-based Marc Howard teamed up with over 500 machine learning 専門家s to crowdsource 感情 分析 algorithms, bringing in data from sources 含むing Google 傾向s, Reddit and 開発 壇・綱領・公約 GitHub.

Howard said his bitcoin 投資s using an algorithm （警官の）巡回区域，受持ち区域 基金s 簡単に 跡をつけるing the price of the cryptocurrency by 54% in the year to June 24, 追加するing that 基金s in New York and Taipei had tapped him for help in developing their own 分析.

"It's pretty hot 権利 now," he said. "Any 基金 that's 価値（がある） their salt, they are 充てるing some of their 資源s and 配分 for 感情 分析."

(報告（する）／憶測ing by Tom Wilson and Simon Jessop; Editing by Pravin Char)

Sorry we are not 現在／一般に 受託するing comments on this article.